# Authentication
Source: https://docs.ayliea.com/api-reference/authentication
Create and manage API keys to authenticate requests to the Ayliea REST API.
All API requests are authenticated using API keys passed in the `X-API-Key` header. API keys are scoped to specific endpoints and tied to your organization.
## Creating an API key
API keys are managed in the Ayliea web app by organization Owners and Admins.
Navigate to **Organization Settings** and select the **API Keys** tab.
Click **Create Key**. Provide a descriptive label (e.g., "Splunk Integration" or "Weekly Report Sync") and select the scopes this key needs.
The full API key is displayed once after creation. Copy it and store it securely — you cannot retrieve it again.
If you lose a key, revoke it and create a new one.
## Key format
API keys use the following format:
```
ayliea_pk_<64 hex characters>
```
The `ayliea_pk_` prefix identifies the key type. The first 8 hex characters after the prefix are stored as a visible identifier so you can tell keys apart in the dashboard.
## Using a key
Pass your API key in the `X-API-Key` header on every request:
```bash theme={null}
curl https://assess.ayliea.com/api/v1/scores \
-H "X-API-Key: ayliea_pk_a1b2c3d4e5f6..."
```
Never include API keys in URLs, query parameters, or client-side code. Use the header exclusively.
## Scopes
Each API key is granted one or more scopes that control which endpoints it can access. A request to an endpoint that requires a scope not assigned to the key returns a `403 Forbidden` response.
| Scope | Grants access to |
| - | - |
| `scores:read` | `GET /api/v1/scores` |
| `recommendations:read` | `GET /api/v1/recommendations` |
| `discovery:read` | `GET /api/v1/discovery` |
| `assessments:read` | `GET /api/v1/assessments` |
Create keys with the minimum scopes needed for each integration. A SIEM that only needs scores should not have `discovery:read` access.
## Key storage and security
API keys are hashed with SHA-256 before storage. Ayliea does not store the raw key — only you have it.
**Best practices:**
* Store keys in a secrets manager (e.g., AWS Secrets Manager, HashiCorp Vault, or your CI/CD platform's secret store)
* Never commit keys to version control
* Use separate keys for separate integrations so you can revoke them independently
* Set the narrowest scopes possible on each key
## Key rotation
There is no automatic key rotation. To rotate a key:
1. Create a new key with the same label and scopes
2. Update your integration to use the new key
3. Verify the integration works with the new key
4. Revoke the old key
Both the old and new keys work simultaneously until the old key is revoked, so you can rotate without downtime.
## Limits
| Constraint | Value |
| - | - |
| Maximum active keys per organization | 25 |
| Key length | `ayliea_pk_` prefix + 64 hex characters (74 characters total) |
Revoking a key frees up one slot immediately. Revoked keys cannot be re-activated.
## Error responses
| Status | Meaning |
| - | - |
| `401 Unauthorized` | Missing `X-API-Key` header or key not found |
| `403 Forbidden` | Key does not have the required scope, or API access is not included in your engagement |
```json theme={null}
// Missing or invalid key
{ "error": "Invalid API key" }
// Wrong scope
{ "error": "API key does not have the 'scores:read' scope" }
// API access not enabled
{ "error": "API access is not enabled for this organization. Contact your account team to enable it." }
```
# Discovery
Source: https://docs.ayliea.com/api-reference/endpoints/discovery
Retrieve the latest AI platform discovery scan results and detected platforms.
```
GET /api/v1/discovery
```
Returns the most recent completed discovery scan for your organization along with the platforms detected during that scan. If no scan has been completed, `scan` is `null` and `platforms` is an empty array.
**Required scope:** `discovery:read`
## Request
```bash curl theme={null}
curl "https://assess.ayliea.com/api/v1/discovery?limit=50" \
-H "X-API-Key: ayliea_pk_YOUR_API_KEY"
```
```python Python theme={null}
import requests
response = requests.get(
"https://assess.ayliea.com/api/v1/discovery",
headers={"X-API-Key": "ayliea_pk_YOUR_API_KEY"},
params={"limit": 50}
)
data = response.json()
scan = data["scan"]
platforms = data["platforms"]
```
```javascript Node.js theme={null}
const response = await fetch(
"https://assess.ayliea.com/api/v1/discovery?limit=50",
{ headers: { "X-API-Key": "ayliea_pk_YOUR_API_KEY" } }
);
const { scan, platforms } = await response.json();
```
## Query parameters
Maximum number of platforms to return. Must be between 1 and 100.
Number of platforms to skip. Must be 0 or greater.
Pagination applies to the `platforms` array only. The `scan` object is always returned in full.
## Response
```json theme={null}
{
"scan": {
"id": "f47ac10b-58cc-4372-a567-0e02b2c3d479",
"organization_id": "550e8400-e29b-41d4-a716-446655440000",
"status": "completed",
"domain": "example.com",
"started_at": "2026-03-15T14:00:00Z",
"completed_at": "2026-03-15T14:30:00Z",
"summary": {
"platforms_discovered": 14,
"shadow_ai_count": 3,
"overall_risk_score": 65
}
},
"platforms": [
{
"id": "c56a4180-65aa-42ec-a945-5fd21dec0538",
"platform_name": "ChatGPT",
"platform_type": "llm_chat",
"url": "https://chat.openai.com",
"confidence": 0.95,
"metadata": {
"vendor": "OpenAI",
"category": "LLM Chat"
},
"created_at": "2026-03-15T14:15:00Z"
},
{
"id": "e7b1f3a0-4c2d-4b8e-9f1a-2d3c4e5f6a7b",
"platform_name": "GitHub Copilot",
"platform_type": "code_assistant",
"url": "https://github.com/features/copilot",
"confidence": 0.88,
"metadata": {
"vendor": "GitHub",
"category": "Code Assistant"
},
"created_at": "2026-03-15T14:18:00Z"
}
]
}
```
## Response fields
The most recent completed discovery scan, or `null` if no scan has been completed.
Unique identifier (UUID) for the scan.
UUID of the organization that owns this scan.
Scan status. Always `completed` in API responses (only completed scans are returned).
The domain that was scanned.
ISO 8601 UTC timestamp of when the scan started.
ISO 8601 UTC timestamp of when the scan finished.
Aggregated scan results.
Total number of AI platforms detected.
Number of detected platforms that are not on the organization's approved list.
Aggregate risk score from 0 to 100 based on the discovered platforms.
Platforms detected during the scan, sorted by confidence score (highest first).
Unique identifier (UUID) for the detected platform record.
Name of the detected platform (e.g., "ChatGPT", "GitHub Copilot").
Classification of the platform (e.g., `llm_chat`, `code_assistant`, `image_generation`).
URL associated with the detected platform.
Detection confidence score from 0.0 to 1.0. Higher values indicate stronger evidence of platform usage.
Additional platform details. Contents vary by platform but typically include `vendor` and `category`.
ISO 8601 UTC timestamp of when this platform was detected.
## No scan completed
If your organization has not completed a discovery scan, the response is:
```json theme={null}
{
"scan": null,
"platforms": []
}
```
## Notes
* Only the most recent completed scan is returned. Historical scans are not available through the API.
* The `metadata` object structure varies by platform and may include additional fields beyond `vendor` and `category` in future versions.
* Platforms are sorted by `confidence` descending. Use pagination to retrieve large result sets.
# Recommendations
Source: https://docs.ayliea.com/api-reference/endpoints/recommendations
Retrieve prioritized remediation recommendations for your organization.
```
GET /api/v1/recommendations
```
Returns recommendations across all completed assessments for your organization, sorted by priority (highest first) and then by creation date (newest first).
**Required scope:** `recommendations:read`
## Request
```bash curl theme={null}
curl "https://assess.ayliea.com/api/v1/recommendations?limit=25" \
-H "X-API-Key: ayliea_pk_YOUR_API_KEY"
```
```python Python theme={null}
import requests
response = requests.get(
"https://assess.ayliea.com/api/v1/recommendations",
headers={"X-API-Key": "ayliea_pk_YOUR_API_KEY"},
params={"limit": 25}
)
recommendations = response.json()["recommendations"]
```
```javascript Node.js theme={null}
const response = await fetch(
"https://assess.ayliea.com/api/v1/recommendations?limit=25",
{ headers: { "X-API-Key": "ayliea_pk_YOUR_API_KEY" } }
);
const { recommendations } = await response.json();
```
## Query parameters
Maximum number of results to return. Must be between 1 and 100.
Number of results to skip. Must be 0 or greater.
## Response
```json theme={null}
{
"recommendations": [
{
"id": "d290f1ee-6c54-4b01-90e6-d701748f0851",
"title": "Enable multi-factor authentication for all users",
"priority": 1,
"status": "open",
"framework_id": "nist-csf-2",
"category": "Access Control"
},
{
"id": "a3bb189e-8bf9-3888-9912-ace4e6543002",
"title": "Implement network segmentation between production and development",
"priority": 2,
"status": "in_progress",
"framework_id": "cis-v8",
"category": "Network Security"
}
]
}
```
## Response fields
Array of recommendation objects.
Unique identifier (UUID) for the recommendation.
Short description of the recommended action.
Priority ranking. Lower numbers indicate higher priority (1 is the highest).
Current status of the recommendation. One of: `open`, `in_progress`, `completed`, `dismissed`.
The framework this recommendation is associated with (e.g., `nist-csf-2`, `hipaa`).
The assessment category this recommendation falls under (e.g., "Access Control", "Network Security").
## Sorting
Results are sorted by:
1. `priority` ascending (highest priority first)
2. `created_at` descending (newest first within the same priority)
This sort order is fixed and cannot be changed via query parameters.
## Notes
* Recommendations span all completed assessments — they are not filtered by framework. To filter client-side, use the `framework_id` field.
* If no assessments have been completed, the `recommendations` array is empty.
* Status changes made in the Ayliea web or mobile app are reflected immediately in API responses.
# Scores
Source: https://docs.ayliea.com/api-reference/endpoints/scores
Retrieve the latest completed assessment scores per framework for your organization.
```
GET /api/v1/scores
```
Returns the most recent completed assessment score for each framework in your organization. If you have completed multiple assessments for the same framework, only the latest result is returned.
**Required scope:** `scores:read`
## Request
```bash curl theme={null}
curl https://assess.ayliea.com/api/v1/scores \
-H "X-API-Key: ayliea_pk_YOUR_API_KEY"
```
```python Python theme={null}
import requests
response = requests.get(
"https://assess.ayliea.com/api/v1/scores",
headers={"X-API-Key": "ayliea_pk_YOUR_API_KEY"}
)
scores = response.json()["scores"]
```
```javascript Node.js theme={null}
const response = await fetch("https://assess.ayliea.com/api/v1/scores", {
headers: { "X-API-Key": "ayliea_pk_YOUR_API_KEY" },
});
const { scores } = await response.json();
```
This endpoint does not accept query parameters. It returns all frameworks with completed assessments.
## Response
```json theme={null}
{
"scores": [
{
"framework_id": "ai-security",
"overall_score": 72,
"grade": "C",
"completed_at": "2026-03-15T14:30:00Z"
},
{
"framework_id": "nist-csf-2",
"overall_score": 85,
"grade": "B",
"completed_at": "2026-03-10T09:15:00Z"
}
]
}
```
## Response fields
Array of score objects, one per framework with a completed assessment.
Framework identifier. One of: `ai-security`, `cis-v8`, `nist-800-53`, `nist-csf-2`, `hipaa`, `iso-27001`, `soc-2`, `pci-dss`.
Weighted aggregate score from 0 to 100.
Letter grade derived from the overall score: `A` (90-100), `B` (80-89), `C` (70-79), `D` (60-69), `F` (0-59).
ISO 8601 UTC timestamp of when the assessment was completed.
## Framework identifiers
| ID | Framework |
| - | - |
| `ai-security` | AI Security |
| `cis-v8` | CIS Controls v8 |
| `nist-800-53` | NIST 800-53 |
| `nist-csf-2` | NIST CSF 2.0 |
| `hipaa` | HIPAA |
| `iso-27001` | ISO 27001 |
| `soc-2` | SOC 2 |
| `pci-dss` | PCI DSS |
## Notes
* Only completed assessments are included. In-progress assessments do not appear.
* If no assessments have been completed, the `scores` array is empty.
* Scores are calculated using weighted category averages. See [Scoring](/product/scoring) for details on how scores are computed.
# Errors
Source: https://docs.ayliea.com/api-reference/errors
HTTP status codes, error response format, and how to handle API errors.
The Ayliea API uses standard HTTP status codes and returns errors in a consistent JSON format.
## Error format
All error responses return a JSON object with a single `error` field containing a human-readable message:
```json theme={null}
{
"error": "Human-readable error description"
}
```
Error messages are intentionally general and do not expose internal implementation details.
## Status codes
| Code | Meaning | When it happens |
| - | - | - |
| `200` | OK | Request succeeded |
| `400` | Bad request | Invalid query parameters (e.g., `limit` out of range) |
| `401` | Unauthorized | Missing or invalid `X-API-Key` header |
| `403` | Forbidden | API key lacks the required scope, or API access is not included in your engagement |
| `404` | Not found | Requested resource does not exist |
| `409` | Conflict | Duplicate resource (e.g., maximum API keys reached) |
| `429` | Too many requests | [Rate limit](/api-reference/rate-limiting) exceeded |
| `500` | Internal server error | Unexpected server-side failure |
## Example error responses
```json theme={null}
{
"error": "Invalid API key"
}
```
```json theme={null}
{
"error": "API key does not have the 'scores:read' scope"
}
```
```json theme={null}
{
"error": "limit must be an integer between 1 and 100"
}
```
```json theme={null}
{
"error": "Too many requests. Please try again later."
}
```
```json theme={null}
{
"error": "Failed to query scores"
}
```
## Handling errors
**`4xx` errors** indicate a problem with your request. Check the error message, fix the issue, and retry.
**`429` errors** mean you have been rate limited. Wait for the `Retry-After` period and retry. See [Rate limiting](/api-reference/rate-limiting) for details.
**`500` errors** indicate a server-side problem. These are safe to retry with exponential backoff. If `500` errors persist, contact [support@ayliea.com](mailto:support@ayliea.com).
# API overview
Source: https://docs.ayliea.com/api-reference/overview
Integrate Ayliea security posture data into your existing tools and workflows with the REST API.
The Ayliea REST API provides programmatic access to your organization's security posture data — assessment scores, recommendations, and AI discovery results. Use it to feed dashboards, trigger alerts in your SIEM, sync remediation tasks with your ticketing system, or build custom integrations.
## Base URL
All API requests use the following base URL:
```
https://assess.ayliea.com/api/v1
```
The API is versioned via the URL path. The current version is `v1`.
## Key details
| Property | Value |
| - | - |
| Protocol | HTTPS only |
| Format | JSON request and response bodies |
| Auth | API key via `X-API-Key` header |
| Availability | Included with an Enterprise engagement — contact your account team to enable API access |
| CORS | Not supported — server-to-server use only |
| Timestamps | ISO 8601 UTC (e.g., `2026-03-15T14:30:00Z`) |
| Rate limit | 60 requests per minute per API key |
The API is designed for server-to-server integration. API keys must never be exposed in client-side code, mobile apps, or browser requests.
## Quick start
Get from zero to your first API call in three steps.
Navigate to **Organization Settings** and open the **API Keys** tab. Click **Create Key**, give it a label (e.g., "SIEM Integration"), select the scopes you need, and click **Create**.
Copy the key immediately — it is shown only once.
Replace `YOUR_API_KEY` with the key you just copied:
```bash theme={null}
curl https://assess.ayliea.com/api/v1/scores \
-H "X-API-Key: ayliea_pk_YOUR_API_KEY"
```
A successful response returns JSON with your latest assessment scores:
```json theme={null}
{
"scores": [
{
"framework_id": "ai-security",
"overall_score": 72,
"grade": "C",
"completed_at": "2026-03-15T14:30:00Z"
}
]
}
```
## Available endpoints
Latest assessment scores per framework with grades and completion timestamps.
Prioritized remediation recommendations with status, framework, and category context.
AI platform discovery scan results with detected platforms and confidence scores.
## Next steps
Learn how API keys work, how scopes are enforced, and how to rotate keys safely.
Understand error responses, HTTP status codes, and how to handle failures.
# Pagination
Source: https://docs.ayliea.com/api-reference/pagination
Use limit and offset parameters to page through large result sets.
Endpoints that return lists support pagination using `limit` and `offset` query parameters.
## Parameters
| Parameter | Type | Default | Constraints | Description |
| - | - | - | - | - |
| `limit` | integer | `100` | 1 to 100 | Maximum number of results to return |
| `offset` | integer | `0` | 0 or greater | Number of results to skip |
## Usage
Append `limit` and `offset` as query parameters:
```bash theme={null}
# First 25 results
curl "https://assess.ayliea.com/api/v1/recommendations?limit=25" \
-H "X-API-Key: ayliea_pk_YOUR_API_KEY"
# Next 25 results
curl "https://assess.ayliea.com/api/v1/recommendations?limit=25&offset=25" \
-H "X-API-Key: ayliea_pk_YOUR_API_KEY"
```
## Paginated endpoints
The following endpoints support pagination:
| Endpoint | Pagination applies to |
| - | - |
| `GET /api/v1/recommendations` | Recommendation list |
| `GET /api/v1/discovery` | Discovered platforms list |
The `GET /api/v1/scores` endpoint returns one score per framework and does not support pagination.
## Validation errors
Invalid pagination parameters return a `400` response:
```json theme={null}
// limit out of range
{ "error": "limit must be an integer between 1 and 100" }
// negative offset
{ "error": "offset must be a non-negative integer" }
```
## Tips
* Start with the default `limit` of 100 and add pagination only if you need smaller pages
* If you receive fewer results than your `limit`, you have reached the end of the list
* Results are ordered by the endpoint's default sort (e.g., recommendations are sorted by priority ascending, then created date descending)
# Rate limiting
Source: https://docs.ayliea.com/api-reference/rate-limiting
Understand API rate limits, response headers, and how to handle throttled requests.
The Ayliea API enforces rate limits to ensure consistent performance for all customers. Limits are applied per API key.
## Limits
| Endpoint category | Limit |
| - | - |
| All read endpoints (`/v1/scores`, `/v1/recommendations`, `/v1/discovery`) | 60 requests per minute |
The rate limit window is a sliding 60-second window. Once you exceed 60 requests in any 60-second period, subsequent requests receive a `429` response until the window resets.
## Rate limit headers
When a request is rate limited, the `429` response includes headers to help you determine when to retry:
| Header | Description |
| - | - |
| `Retry-After` | Seconds until the current window resets |
| `X-RateLimit-Limit` | Maximum requests allowed per window |
| `X-RateLimit-Remaining` | Requests remaining in the current window (always `0` on a 429) |
| `X-RateLimit-Reset` | Unix timestamp (seconds) when the window resets |
## Example `429` response
```bash theme={null}
HTTP/1.1 429 Too Many Requests
Retry-After: 23
X-RateLimit-Limit: 60
X-RateLimit-Remaining: 0
X-RateLimit-Reset: 1742054423
Content-Type: application/json
```
```json theme={null}
{
"error": "Too many requests. Please try again later."
}
```
## Handling rate limits
When you receive a `429`, wait the number of seconds specified in `Retry-After` before retrying. Do not retry immediately.
If you are making many requests in sequence, add exponential backoff with jitter. Start with a 1-second delay, double it on each retry, and add a random component to avoid thundering herd effects.
```python theme={null}
import time, random
def request_with_backoff(url, headers, max_retries=5):
for attempt in range(max_retries):
response = requests.get(url, headers=headers)
if response.status_code != 429:
return response
wait = (2 ** attempt) + random.uniform(0, 1)
time.sleep(wait)
raise Exception("Max retries exceeded")
```
If you consistently hit rate limits, consider caching responses on your side. Assessment scores and recommendations change infrequently — polling once every 5-15 minutes is sufficient for most integrations.
# 2026 Changelog
Source: https://docs.ayliea.com/changelog/2026
All product updates and new features shipped in 2026.
# 2026 Changelog
## July 2026
### Engagement-based pricing and access
Ayliea has moved from self-serve subscription tiers to expert-led engagements: fixed-scope assessments delivered by a named assessor, with published one-time floors (Focused, Comprehensive, Enterprise) plus an optional Monitoring Retainer to keep your posture and portal access current between assessments. The AI Security (AISS) self-assessment stays free for everyone, no account required. See [Introduction](/introduction) for current pricing and [Frameworks Overview](/frameworks/overview) for what's included at each stage.
### Accounts are now provisioned, not self-signed-up
Public sign-up has been replaced with admin-provisioned access: your organization and account are set up by Ayliea as part of starting an engagement, and you activate your own account from a secure emailed link — set your password, enroll multi-factor authentication, and save your backup codes. See [Getting Access](/getting-started/creating-account) for the full activation walkthrough.
### Engagement Deliverables portal
A new **Deliverables** page in your portal shows the consulting artifacts and report snapshots your assessor publishes as your engagement progresses, grouped by engagement with a "Delivered" badge once your assessor marks the engagement complete. See [Deliverables](/product/deliverables).
### Engagement and retainer invoices in your portal
A new **Invoices** page shows invoices for your engagement and any Monitoring Retainer, with status (open, paid, void, past due) and online payment for open invoices via a secure hosted payment page. See [Billing & Invoices](/product/billing).
## June 2026
### AI Governance Report
Generate an audit-ready AI governance report for your organization on demand. The report brings together your AI systems inventory, your compliance posture across the frameworks you've activated, and the regulatory milestones ahead of you — including EU AI Act enforcement dates — into a single, professionally formatted PDF for auditors, your board, or internal governance reviews. Available on the Business plan and above.
### Cyber-insurance underwriting crosswalk
Turn your security and AI-governance posture into the documentation cyber-insurance underwriters ask for. Choose your insurer profile, preview how your current evidence maps to common underwriting questions, then export a submittable crosswalk — so renewal and new-policy questionnaires start from your real posture instead of a blank form. Available on the Business plan and above.
### AI Acceptable Use Policy generator
Generate a starting AI Acceptable Use Policy for your organization as an editable Word document, drawn from your AI Security assessment. The draft reflects what your assessment indicates about how your organization uses and governs AI, and ships with a clear disclaimer so you can review, tailor, and adopt it as your own — a head start on a governance document instead of a blank page. Available from your AI Governance dashboard.
## May 2026
### AI Vendor Watch in your sidebar (2026-05-29)
The AI Vendor Watch change feed is now reachable from the sidebar, the mobile More menu, and the AI Governance dashboard — not just by direct URL. When sub-processors, data residency, or certifications change on the AI vendors you depend on, the changes surface here, and critical and high-severity changes also trigger email alerts to the org owner.
The change feed itself rolled out earlier in May; this update closes the discoverability gap so you can actually find it in normal navigation. Available on the Business plan and above.
### ✅ AI Autofill is back — and starts automatically (2026-05-27)
AI Questionnaire Autofill is live again on the Business plan and above, after the hardening work from the May 21 pause shipped. What's new since then:
* **Uploads parse automatically.** Drop in a questionnaire (PDF, DOCX, or CSV) and it's split into individual questions and ready to draft in the same step — no extra click.
* **Drafts ground in more of your evidence.** Answers now draw on your assessment responses, connected-tool evidence, and uploaded files — not just framework reference text — and every answer still carries a citation chain. Questions without enough supporting evidence are flagged for review, never fabricated.
* **Easier to find.** Reach it from the sidebar and the AI Governance dashboard, not just a direct link.
As before, your questionnaire content and evidence are never used to train any AI model.
### ⚠️ AI Autofill paused for new uploads (2026-05-21)
Trust is the entire point of the AI Questionnaire Autofill product. A production-readiness audit on 2026-05-21 found gaps between what the feature claimed and what the implementation guaranteed: data-handling specifics, prompt-injection defenses, citation integrity, stuck-state recovery, and per-organization cost controls all needed work before the feature could responsibly meet SMB security buyers.
Rather than ship marketing claims we couldn't back end-to-end, we paused new uploads and new draft generation. Questionnaires already in your account remain reviewable, finalize-able, and exportable. The product returns when it meets our own bar.
For status updates, email [support@ayliea.com](mailto:support@ayliea.com).
### NIST AI 600-1 (Generative AI Profile) framework
Ayliea now ships the NIST AI RMF Generative AI Profile (NIST AI 600-1, July 2024) as a standalone assessment. The profile organizes the 12 generative-AI-specific risks from §2 — CBRN information, confabulation, dangerous or hateful content, data privacy, environmental impact, harmful bias and homogenization, human-AI configuration, information integrity, information security, intellectual property, obscene or abusive content, and value chain integration — and maps each one against the AI RMF GOVERN, MAP, MEASURE, and MANAGE functions from §3. Forty-four questions total, source-honest: Environmental Impacts has only a MEASURE question because the published §3 lists no other actions for that risk, and Confabulation has no MAP question for the same reason. Every question's remediation playbook cites the specific AI 600-1 Action IDs (GV-1.3-002, MS-2.6-005, and the like) that apply, so audit-ready evidence ties back to the published document. Crosswalk references to the parent NIST AI RMF, ISO/IEC 42001, OWASP LLM Top 10, and EU AI Act high-risk requirements ship with every recommendation. Especially useful for healthcare and financial-services teams subject to sector oversight, US contractors under EO 14110, and organizations integrating third-party foundation models.
### AI Questionnaire Autofill
Upload a security questionnaire (PDF, DOCX, or CSV) from a customer, auditor, or prospect and Ayliea drafts an answer for every question, grounded in your organization's existing evidence and prior responses. Every draft carries a citation chain — you can see exactly which sources backed each answer, plus a confidence score — and questions without sufficient evidence are flagged inline rather than guessed. Review, edit, or reject drafts in the reviewer UI, then export the finalized questionnaire as CSV for spreadsheet review, JSON for system integration, or DOCX for an auditor-ready document. Available on the Business plan and above.
### AISS framework — expanded to 66 sub-controls
AISS now covers 66 sub-controls across the same 10 control domains. Twelve sub-controls — AI Risk Assessment, Executive Sponsorship, AI Suitability Criteria, Embedded AI Discovery, PII in AI Systems, AI Integration Controls, AI Vendor Monitoring, AI Cost Monitoring, AI Guidance Documentation, AI Training Effectiveness, Synthetic Content Provenance, and Agentic AI Action Authorization — are now formally part of the standard your assessment is scored against, each with a corresponding remediation playbook. The expanded coverage closes gaps practitioners and auditors had asked about, particularly around embedded AI (Copilot-in-Office), agentic AI workloads, and synthetic content provenance.
### AISS remediation playbooks — full coverage
Every AISS sub-control (66 controls in total) now has a remediation playbook. AI-generated recommendations for AISS assessments now include an objective, the risk if the control is missing, three concrete remediation steps, environment-specific guidance for common stacks, and cross-framework references to NIST CSF 2.0, NIST AI RMF, CIS Controls v8.1, ISO/IEC 27001:2022, the EU AI Act, and the Colorado AI Act. Recommendations are no longer generic — every AISS gap your assessment surfaces now comes with auditor-ready remediation depth.
### Vertical-aware AI recommendations
When your organization selects a vertical default (Healthcare or Financial Services), the AI-generated recommendations on every assessment now speak to the regulatory and threat surface for your sector. Healthcare orgs see recommendations biased toward PHI handling in inference paths, business associate agreements, and breach notification timelines. Financial Services orgs see recommendations biased toward model validation cadence, MNPI minimization, and adversarial robustness for fraud and trading models. The personalized summary on every recommendation now says "in the Healthcare sector" or "in the Financial Services sector" rather than free-form industry text. If no vertical is set, recommendations continue to use your free-form industry — there is no change for non-vertical organizations.
### AISS Coverage hub at `/aiss`
Every authenticated user with an AISS assessment now has a dedicated coverage page showing posture across all 10 AISS control domains (AC-1 through AC-10) in a single grid. Each domain card surfaces the score and links directly to the corresponding domain markdown in the public [Ayliea/aiss](https://github.com/Ayliea/aiss) repository — so auditors can verify your score against the published spec without leaving the page. Available from the sidebar (desktop) and the More sheet (mobile).
### Glass-Box Score breakdown on the results page
For AISS assessments, the score breakdown is now expandable per category. Each AC-N domain drops down to show the questions answered, weight per question, points earned vs. maximum, MITRE ATLAS technique mappings, framework crosswalks (NIST CSF 2.0, ISO/IEC 27001:2022, NIST AI RMF, CIS Controls v8.1, EU AI Act, Colorado AI Act), and a deep-link to the matching domain markdown in the public AISS repository. Any auditor can hand-calculate the score from the published spec plus your answers.
### In-app RFC deep-link from disputed sub-controls
Every question row in the Glass-Box audit drilldown now has a pencil icon that opens a pre-filled GitHub issue on [Ayliea/aiss](https://github.com/Ayliea/aiss) with the control ID, question text, selected option, and points earned populated. Practitioners who disagree with how a control is scored have a one-click path to propose a change.
### AISS published as an open standard
The Ayliea AI Security Standard is now publicly available at [github.com/Ayliea/aiss](https://github.com/Ayliea/aiss) under [CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/). 10 control domains, 56 sub-controls, 9 framework crosswalks (NIST CSF 2.0, ISO/IEC 27001:2022, NIST AI RMF, NIST AI 600-1, CIS Controls v8.1, EU AI Act, Colorado AI Act, OWASP LLM Top 10, MITRE ATLAS). Fork it, audit your organization with it, or propose changes through the public RFC issue template — the same standard your in-app Ayliea score is computed against. Current production tag: `v1.2.3`.
### Vertical threat profiles — Healthcare, Financial Services, Legal
Three industry pages now overlay MITRE ATLAS techniques with the threats specific to each vertical's AI workloads. Healthcare anchors on patient-data exfiltration via clinical-decision-support agents. Financial Services covers prompt-injection-driven fraud in customer-facing chat and adversarial drift in credit/risk models. Legal covers the privilege-erosion risks of LLM-assisted document review.
### Vertical compliance bundles — Healthcare and Financial Services
Both verticals now ship a coordinated bundle page that maps AISS against the vertical's required frameworks. Healthcare bundles HIPAA Security Rule + FDA SaMD guidance + AISS. Financial Services bundles SOC 2 + PCI DSS + AISS with model-risk-management overlays for NYDFS Part 500, EU DORA, SR 11-7, FINRA, and SEC Marketing Rule. Buyers in regulated industries can self-qualify Ayliea's coverage against their compliance reality in one page.
### PLG pricing restored
Pricing is back to four transparent tiers: Free ($0), Pro ($500/yr), Business ($2,500/yr), and Enterprise (published floor — starting at $15,000/yr). Pro and Business sign up directly and pay via Stripe Checkout. Enterprise is inbound only — visit `/contact` to book a tailored demo. Replaces the 2026-05-04 sales-led pricing inquiry form.
*Update (2026-05-12): Pro and Business were re-priced the following day to **$1,200/yr** and **$3,600/yr** respectively after competitor validation. The $500 / $2,500 numbers above record what shipped 2026-05-11 only — see [/pricing](/pricing) for current rates.*
### Self-serve sign-up restored
Public account creation is back at `/sign-up`. NIST SP 800-63B password validation (12-character minimum), tier-aware copy on Pro and Business signup flows, and paid-tier signups proceed directly to checkout. Replaces the redirect-to-sign-in route from the sales-led pivot.
### Account activation flow
Newly invited members complete onboarding through a single `/activate` page that walks them through password setup (NIST SP 800-63B compliant — 12-character minimum with compromised-password check), MFA enrolment via TOTP, and backup-code download. Whether you're an Account Owner activating your org for the first time or a teammate joining an existing org, the flow is identical.
### Account Owner team management
Owners and admins can now invite teammates by email, resend activation emails to members who haven't signed in yet, change roles with full audit-log trail, and remove members. New-to-Ayliea invitees get an auth account provisioned automatically and an activation email; existing Ayliea users get an org-membership invitation. When a removed member has no remaining org memberships, their auth account is fully deleted.
### Suspend members
Owners and admins can suspend a member, which revokes all active sessions immediately and denies any subsequent re-auth attempt. Unsuspend restores access on the member's next sign-in.
### Member observability
The team list now shows status (Invited / Active / Suspended), MFA enrolment, and last-sign-in time per member. Lists are paginated for orgs with more than 50 members.
### Per-organization feature gating
Every org now carries a bundle assignment (Free / Core / Discover +) plus per-feature overrides on top of the bundle's defaults. Sales-led deals can enable specific capabilities per customer without forcing them into a higher tier.
### Public Trust Center at /trust
Documents Ayliea's encryption (per-organization envelope encryption with DEK + MEK), authentication (TOTP MFA mandatory, NIST 800-63B password policy), audit logging, and sub-processors. Enterprise buyers can self-serve the security questions they previously asked on every demo call.
### Trust Center scannability redesign
The `/trust` page now opens with a 4-card fast-facts header (Encryption, MFA, Sub-processors, DPA) so buyers can answer the top security questions in under 5 seconds. A sticky table of contents pins to the left rail on desktop and highlights the section currently in view; a Jump-to-section accordion replaces it on mobile. A "Last verified" timestamp under the hero makes the page's freshness explicit.
### Trust Center: regulatory alignment, not certification
The GDPR and CCPA entries on `/trust` are now labelled "Aligned" rather than "Compliant", with notes that describe what we actually do operationally (DPA published with SCCs Module Two for international transfers, data subject rights honored via the privacy notice). Both are self-assessed regulatory standards with no third-party certification — the previous framing implied a higher bar than we could defend without external legal review. The section title also changed from "Certifications & attestations" to "Regulatory alignment".
### Competitor comparisons, glossary, vendor lookup, AI Risk Score Calculator
New marketing surfaces: per-competitor comparison pages at `/compare/`, per-term definitions at `/glossary/`, a free vendor risk-classification tool at `/tools/vendor-lookup`, and an AI risk scoring calculator at `/tools/ai-risk-score`.
### Sales-led pricing
Public tier names and dollar amounts have been removed; pricing is now sales-led through `/contact`. The in-app `/upgrade` page is now a read-only bundle browser.
### Public self-signup removed
`/sign-up` is gone. New customer accounts are created exclusively by Ayliea admins (via the provisioning UI) and Account Owners (via the team-management flow).
### Email deliverability hardening
Apex SPF record now correctly authorizes Google Workspace; Resend DKIM record carries the standard tag prefix. mail-tester.com scores 10/10 for outbound mail from both Workspace and Resend transactional paths.
## April 2026
### Continuous monitoring drift alerts (Business tier)
When evidence from your connected integrations causes your assessment scores to drop, Ayliea now detects the drift and emails your organization owner with a breakdown of the affected controls, their previous and current scores, and a deeplink back to the assessment. The continuous-monitoring sweep runs nightly — no configuration needed, just connect an evidence source and you're covered. Available on Business and above.
### Scheduled GitHub evidence polling
Your connected GitHub integration now refreshes evidence automatically every morning — no more needing to hit "Refresh" manually. The scheduled poll covers your most-recent completed assessment across up to three frameworks per organization, so a Business-tier team running both CIS v8 and SOC 2 sees both updated. Combined with drift alerts, this closes the loop: your posture score now tracks the real-world state of your tooling on a rolling basis.
### Drift alert severity
Drift emails now carry a severity tier. Critical regressions (dropping below passing or losing more than 15 points overall) get a red banner and a `[CRITICAL]` subject prefix so your team can triage at a glance. Lighter regressions stay amber.
### Score trend chart on the results page (Business tier)
Every assessment now shows a line chart of how your score has moved over time. Points are colored by what caused each rescore — completion (your initial baseline), evidence poll (automatic updates from your connected integrations), or manual rescore. The card lives on the assessment results page and appears as soon as you have a snapshot, then becomes a proper line once a second snapshot lands. A delta indicator shows the change since the previous capture, so you can see at a glance whether you're trending toward or away from your posture target.
### HIPAA Security Rule Assessment
Assess your organization against the full HIPAA Security Rule — 80 questions covering Administrative, Physical, Technical, Organizational, and Documentation safeguards. Results are scored against the five sections of 45 CFR Part 164 Subpart C, with actionable remediation guidance for each gap. Available on Pro and above.
### SOC 2 Type II Security Assessment
Measure your posture against the AICPA Trust Services Criteria 2017 (revised 2022) — 84 questions covering all nine Common Criteria (CC1 Control Environment through CC9 Risk Mitigation). Map evidence to the controls auditors actually ask about. Available on Pro and above.
### AI Agent Security Framework
A focused mini-framework for teams deploying AI agents in production — 26 questions across six categories: agent governance, delegated authority and credentials, tool invocation security, context and memory protection, monitoring and incident response, and multi-agent orchestration. Aligned to MITRE ATLAS v5.1 agentic techniques. Available on Free.
### AI Agent Security assessment fixed for Free tier
A recent issue prevented Free-tier accounts from starting the AI Agent Security assessment. This is now resolved — all Free accounts can run the 26-question assessment without restriction.
### NIST IR 8401 — Satellite Ground Segment Assessment
Applies the NIST Cybersecurity Framework structure to commercial satellite command-and-control systems. 82 questions across six categories (asset management, governance/risk/supply chain, access control and data security, awareness and protective technology, anomaly detection and continuous monitoring, incident response and recovery). For space systems operators and ground segment providers. Available on Pro and above.
### ISO/IEC 42001:2023 AI Management System Assessment
The first certifiable AI management system standard — now fully supported. 69 questions provide complete coverage of clauses 4–10 (context, leadership, planning, support, operation, performance evaluation, improvement) and all 38 Annex A reference controls across A.2 (AI Policies), A.3 (Internal Organization), A.4 (Resources), A.5 (Impact Assessment), A.6 (AI System Life Cycle), A.7 (Data), A.8 (Information for Interested Parties), A.9 (Use of AI Systems), and A.10 (Third-Party and Customer Relationships). Every question cites its exact ISO clause or Annex A control reference so assessments map directly to an external audit. Available on Pro and above.
### OWASP Top 10 for LLM Applications (2025)
The canonical industry reference for LLM application security — now fully assessable in Ayliea. 77 questions across all 10 OWASP categories: prompt injection, sensitive information disclosure, supply chain, data and model poisoning, improper output handling, excessive agency, system prompt leakage, vector and embedding weaknesses, misinformation, and unbounded consumption. Every OWASP-listed prevention strategy maps to a question in its OWASP-listed category, so an LLM10 score reflects coverage of all 15 LLM10 strategies, not a curated subset. Each question includes a remediation playbook authored from the official OWASP guidance, with concrete steps, environment guidance for small/mid/enterprise, and cross-framework references to AISS, ISO 42001, NIST AI RMF, and MITRE ATLAS techniques. Source: [genai.owasp.org/llm-top-10/](https://genai.owasp.org/llm-top-10/) (2025 release, March 12, 2025). Available on Pro and above.
### Personalized remediation for ISO/IEC 42001 findings
Every ISO/IEC 42001 control — all 69 — now has an AI-powered remediation playbook authored from the normative Annex B implementation guidance. When you score a low maturity level on any ISO 42001 control, the recommendation engine personalizes concrete next steps to your organization's size, sector, and environment, so your gap report reads as a prioritized to-do list rather than an abstract control reference.
### Trust Centers
Publish a branded public trust page at `ayliea.com/trust/` showing your framework scores, compliance coverage, and certification badges. Choose between Summary (grades only), Standard (numeric scores), or Detailed (per-category breakdown) visibility. Optionally display an evidence indicator showing how many controls have supporting evidence on file — file contents and names are never exposed. Pro and above for basic trust centers; Business and above for advanced customization.
### AI Usage Policy Engine
Define policies for every AI tool in your organization: Approved, Monitor, or Restricted. Ayliea automatically checks discovery scan results against your policies and flags violations with severity scoring. Export blocklists in CSV, JSON, or text format for direct import into Zscaler, Netskope, or Palo Alto firewalls.
### Policy Violation Trends
Track policy violations over time with a stacked area chart showing severity breakdown (critical, high, medium, low). View resolution rates, current open violations, and select between 7-day, 30-day, and 90-day time ranges.
### Public REST API
Enterprise-tier organizations can now access assessment scores, recommendations, and discovery results programmatically via the v1 REST API. API keys are scoped, hashed at rest, and rate-limited. Full documentation available at [docs.ayliea.com/api-reference](/api-reference/overview).
### Third-Party Integrations
Connect Ayliea to your existing workflow tools. Push recommendations directly into Jira, Linear, GitHub Issues, or Azure DevOps as actionable tickets. Connect Slack for real-time notifications on assessment completions and discovery alerts.
### Webhook Notifications
Subscribe to platform events (assessment completed, recommendation updated, discovery completed, policy violation detected) and receive HMAC-signed webhook deliveries to your endpoints. Track delivery status and retry failed deliveries.
### Scheduled Discovery Scans
Organizations with continuous monitoring enabled now receive automated daily discovery scans. New AI tools and risk escalations trigger alerts automatically without manual intervention.
### Scheduled Report Delivery
Configure weekly or monthly security digest emails delivered to your inbox. Reports include framework scores, recommendation progress, stale assessment warnings, and overdue items.
### Industry Benchmarking
Compare your security scores against anonymized industry benchmarks. See where you stand relative to the 25th, 50th, and 75th percentiles for your industry and framework.
### AI-Powered Recommendations
Recommendations are now personalized by Claude AI based on your organization's size, industry, device environment, and platform stack. The AI tailors remediation steps to reference the specific tools, admin consoles, and settings paths relevant to your setup.
### Account Security Improvements
* Password changes now require current password re-authentication
* Concurrent session limiting: one active session per user, with forced sign-out on other devices
* Account deletion uses a 30-day soft-delete recovery window with email notification
* GDPR data export: download all your personal data in JSON format from Security settings
### Collaborator names and emails encrypted at rest
When you invite an external collaborator to answer a scoped section of an assessment, the name and email you provide them are now protected at rest with the same per-organization encryption used for other personal information across Ayliea. The plaintext is decrypted only when you view the link in your organization page or when the collaborator opens their personal answer link — it never lives in storage in readable form.
### Billing Simplification
Removed per-seat pricing. All tiers now include generous fixed seat limits: Free (1), Pro (10), Business (25), Enterprise (unlimited). No more seat selectors, add-ons, or proration complexity.
## March 2026
### Network Discovery
Upload firewall or proxy logs to automatically discover AI tools active on your network. Ayliea detects shadow AI, scores risk per platform, and maps data flows — giving you visibility into AI usage you may not have sanctioned or even known about.
### AI Tool Inventory
A consolidated view of all discovered AI platforms across your organization. Filter by risk level, approval status, or data access scope, and export your full inventory as a CSV for auditors or procurement reviews.
### Continuous Monitoring
Stream firewall logs via syslog for always-on AI tool detection. Automated alerts notify your team when new shadow AI appears on the network, so you can respond before sensitive data leaves your environment.
### Collector Agent
A lightweight agent for network-level AI tool detection. Privacy-first — it analyzes traffic metadata only. No content decryption, no payload inspection. Deployable in minutes alongside your existing network infrastructure.
### Trust Gap and Verified Score
See the gap between your self-reported assessment score and what network discovery actually observes. The Trust Gap highlights where your stated controls do not match your operational reality, surfacing the highest-priority items for immediate attention.
### Executive Dashboard
A unified view combining your assessment posture with your network discovery findings. The Trust Gap anchors the dashboard, giving security leaders a single screen that shows both what you report and what the network sees.
### PDF Posture Report
Generate a 7-page branded report combining assessment results and discovery data. Includes an executive summary, category score breakdown, AI tool inventory, and a prioritized remediation roadmap. Formatted for auditors, executives, and board presentations.
### Email-Based Invitations
Invite team members to your organization by email with role assignment at the time of invitation. Invitees receive a secure link and are automatically added to your organization after completing sign-up.
### Auditor Report Sharing
Share security posture reports with external auditors via secure, time-limited links. Links require email verification and support configurable scope so you control exactly what the recipient can see.
### Simplified Role System
Streamlined organization roles from five to three: Owner, Admin, and Member. External auditors now access reports through secure share links rather than organization seats, making it easier to manage your team while keeping auditor access appropriately scoped.
# AI Agent Security
Source: https://docs.ayliea.com/frameworks/ai-agent-security
Controls for AI agents acting with delegated authority — least-privilege tool grants, blast-radius limits, prompt-injection defense, and full audit trails.
# AI Agent Security
AI agents are LLMs given the ability to take actions in the real world — call APIs, write to databases, send emails, execute code, browse systems. The blast radius of an agent equals the blast radius of every tool it can call. This framework covers the controls that make agentic deployments safe.
The assessment is appropriate when your organization runs customer-facing copilots, internal automation agents, computer-use agents (e.g., browser-driving assistants), code-writing agents, or any LLM-driven workflow that takes consequential actions beyond returning text.
Use AI Agent Security alongside AI Security (AISS) for organizations deploying agents at scale. AISS covers the broader AI program; AI Agent Security drills into the specific guardrails required for agentic systems.
## What this framework covers
The assessment spans control areas that directly address the OWASP LLM06 Excessive Agency risk and the MITRE ATLAS agentic-AI technique cluster (T0080, T0086, T0098–T0112).
Approval workflow before agents reach production, risk classification of agent use cases, complete agent inventory, deployment controls, and pre-production review processes.
OAuth scope limits, API key scoping and rotation, user-impersonation boundaries, secret storage patterns, and the credential model for agents acting on a user's behalf.
Per-agent tool manifests, human-in-the-loop approval gates for irreversible actions, server-side blast-radius limits (rate caps, monetary ceilings, scope predicates), and write-action authorization.
Defense against direct and indirect prompt injection: input validation, retrieved-content trust boundaries, instruction-data separation, and output filtering before downstream actions execute.
Per-invocation audit logs tying actions to both the agent and the human principal, anomaly detection on tool-call patterns, prompt/response logging, and tamper-evident log retention.
## Why this matters for customers
Customers deploying AI agents face a structural risk that didn't exist with traditional software: one prompt injection or one misaligned decision can cascade into data loss, financial loss, or customer-facing incidents. The AI Agent Security assessment surfaces:
* Whether your agents have least-privilege grants or whether they inherit broad user permissions
* Whether irreversible actions (deletes, payments, account changes, code merges, outbound customer messages) require a synchronous human approval gate
* Whether tool invocations are logged with sufficient detail to investigate after a guardrail failure
* Whether your team has run adversarial testing on the agent guardrails specifically
## How it relates to other frameworks
AI Agent Security focuses narrowly on the agent layer. For the surrounding controls — data governance, model security, supply chain, training and awareness — pair it with the **AI Security (AISS)** assessment. For LLM application risks more broadly (not just agents), see the **OWASP LLM Top 10** assessment. For the foundational risk-management lifecycle, see **NIST AI RMF**.
## Glass-Box scoring
Every score on this framework is reproducible from the published rubric: weighted category scores, per-question maturity tiers, and citations to MITRE ATLAS technique IDs and OWASP LLM Top 10 entries. Auditors can verify any sub-score line by line.
# AI Security Assessment (AISS)
Source: https://docs.ayliea.com/frameworks/ai-security
Score your AI surface against the open Ayliea AI Security Standard — 10 control domains, 59 sub-controls, 6 framework crosswalks, published under CC-BY-4.0.
# AI Security Assessment (AISS)
AISS — the **Ayliea AI Security Standard** — is the open methodology behind every AI Security assessment on the platform. The full specification is published at [github.com/Ayliea/aiss](https://github.com/Ayliea/aiss) under the [Creative Commons Attribution 4.0 license](https://creativecommons.org/licenses/by/4.0/), so you can verify any score from the spec, fork the standard for your own internal use, or propose changes through the public RFC process.
The current production version is **v1.3.0**: 10 control domains, 59 sub-controls, and formal crosswalks to NIST CSF 2.0, NIST AI RMF, CIS Controls v8.1, ISO/IEC 27001:2022, the EU AI Act, and the Colorado AI Act, plus technique-level mappings to MITRE ATLAS.
AISS is the only assessment available without an engagement — any organization can run it at no cost.
## Why an open standard
Most AI security scoring is opaque — proprietary algorithms with no way for an auditor to reproduce the math. AISS takes the opposite position: every score is fully derivable from the answers you provide and the published spec. Anyone with the standard and your answers can recompute your score line-by-line.
That transparency matters in three places:
* **In the audit room** — hand auditors the published JSON spec; they verify framework crosswalks, scoring rubric, and sub-control citations against their own reference frameworks.
* **At the boardroom** — board-ready reports cite a standard your board's outside counsel can read directly.
* **In your security program** — fork AISS internally, extend it for your environment, or propose upstream changes through the public RFC process.
## The 10 control domains
The framework spans 87 questions across 10 AI-specific control domains. Each domain maps to one or more clauses in the source frameworks above.
Organizational policies, roles, and accountability structures for AI adoption and oversight. Includes acceptable-use policies, executive AI literacy, AI ethics committees, and the connection between AI use and the broader risk-management program.
Discovery, inventory, classification, and lifecycle management for every AI tool, model, agent, and integration in your environment. Covers AI development tool governance (Copilot, Cursor, etc.) and shadow-AI detection.
Controls for data flowing to, from, and within AI systems. Classification, encryption, minimization, retention, RAG and vector-store security, and synthetic-content provenance.
Authentication, authorization, privileged-access management, and least-privilege access to AI tools and services. Covers human-to-AI, AI-to-system, and agent-to-agent authorization.
Vendor assessment, contract review, model provenance, and continuity planning for third-party AI services. Includes assessing model origin, training-data lineage, and sub-processor exposure.
Human oversight, accuracy monitoring, bias detection, content disclosure, automated quality controls, and synthetic-content marking (per EU AI Act Article 50 and similar deepfake-disclosure laws).
Incident planning, reporting, classification, containment, post-incident review, and regulatory notification for AI-specific events — including prompt-injection attacks, data leakage via AI output, model compromise, and agent guardrail failures.
Usage logging, anomaly detection, prompt/response logging, performance monitoring, and log integrity for AI systems. Connects AI activity to your existing SIEM and incident-response workflows.
Security awareness programs, role-based training, acceptable-use communication, threat awareness (phishing-via-AI, social engineering of AI agents), and executive AI literacy.
Model access controls, adversarial testing, versioning, integrity verification, prompt-injection defense, and agentic AI guardrails (per MITRE ATLAS T0080–T0112 and OWASP LLM06 Excessive Agency).
## Glass-Box Score breakdown
For every AISS assessment, the results page includes a per-domain expandable breakdown. Each AC-N panel shows the questions answered, weight per question, points earned vs. maximum, MITRE ATLAS technique mappings, framework crosswalks, and a deep-link to the matching domain markdown in the public [Ayliea/aiss](https://github.com/Ayliea/aiss) repository. Any auditor can hand-calculate the score from the published spec plus your answers — no proprietary algorithm involved.
A separate **AISS Coverage** page at `/aiss` rolls all 10 domains into a single posture grid so you can see your overall AI security posture in one view, with each domain card linking directly to the corresponding spec page.
## Propose a change
If a question or scoring rubric doesn't match your environment's reality, every question row in the Glass-Box drilldown has a pencil icon that opens a pre-filled GitHub issue against [Ayliea/aiss](https://github.com/Ayliea/aiss). The RFC issue template captures the control ID, question text, your selected option, and points earned automatically — practitioners shape the standard through the same public process they use to consume it.
# CIS Controls v8.1
Source: https://docs.ayliea.com/frameworks/cis-v8
A prioritized set of 18 security controls published by the Center for Internet Security, designed to give organizations a practical baseline for cyber defense.
# CIS Controls v8.1
The CIS Controls (formerly CIS Critical Security Controls) are published by the Center for Internet Security and represent a curated set of actions that, when implemented, defend against the most prevalent and damaging cyber attacks. Version 8.1 is the current release, updated to reflect modern IT environments including cloud workloads, remote work, and mobile devices.
What sets CIS apart from larger federal frameworks is its emphasis on prioritization. Controls are organized into three Implementation Groups (IG1, IG2, IG3), each building on the last. IG1 represents the minimum standard of information security — sometimes called "essential cyber hygiene" — and is recommended for all organizations regardless of size or industry. IG2 and IG3 extend coverage for organizations with greater resources and higher risk profiles.
With 156 questions across 18 control areas, the CIS v8.1 assessment is well-suited for small and medium-sized organizations wanting a structured, actionable security baseline, as well as larger enterprises using CIS as a complementary framework alongside NIST or ISO 27001.
## What Ayliea Assesses
The assessment maps to all 18 CIS Controls, grouped by the major domains they address. Questions evaluate both the presence of controls and their maturity — whether they are documented, tested, and monitored.
Whether your organization maintains accurate inventories of enterprise hardware and software assets. Questions cover discovery processes, unauthorized device detection, and software allowlisting.
How you classify, handle, and protect data throughout its lifecycle — at rest, in transit, and at disposal. Includes data flow mapping and encryption coverage.
The state of your configuration management for enterprise assets and software, as well as account management practices, access control policies, and privileged access governance.
Your vulnerability management program maturity — scanning frequency, remediation SLAs, and prioritization — alongside audit log configuration, retention, and review processes.
Technical controls protecting against phishing, malicious email attachments, malicious web content, and malware execution. Includes email filtering, browser security, and endpoint protection.
Network monitoring and defense capabilities — segmentation, firewall policies, traffic filtering — as well as your data recovery program, backup coverage, and restoration testing.
The breadth and effectiveness of your security awareness training program, penetration testing and red team activity, application security practices, and incident response plan maturity.
Due diligence applied to third-party service providers, contract requirements, and ongoing vendor security monitoring. Also covers internal penetration testing scope and frequency.
## Who Needs This Assessment
* Small and medium-sized businesses building their first structured security program
* Organizations that want a practical, prioritized security baseline before pursuing heavier frameworks
* Security teams using CIS as a foundational layer alongside NIST CSF or ISO 27001
* Organizations preparing for cyber insurance renewals that reference CIS Controls
* IT teams benchmarking against industry-standard security hygiene
* Organizations in any industry that do not yet have a formal compliance requirement but want to reduce cyber risk systematically
## Getting Started
The CIS Controls v8.1 assessment requires an Organization subscription. Once subscribed, select **CIS Controls v8.1** from the framework list in the Assess tab and begin with whichever control area is most relevant to your current priorities.
For step-by-step instructions, see the [Quickstart guide](/getting-started/quickstart).
# HIPAA Security Rule
Source: https://docs.ayliea.com/frameworks/hipaa
Assessment covering the HIPAA Security Rule requirements for protecting electronic protected health information, required for healthcare organizations and their business associates.
# HIPAA Security Rule
The Health Insurance Portability and Accountability Act (HIPAA) Security Rule, published by the U.S. Department of Health and Human Services (HHS), establishes national standards for protecting electronic protected health information (ePHI). It applies to covered entities — healthcare providers, health plans, and healthcare clearinghouses — as well as any business associates that create, receive, maintain, or transmit ePHI on their behalf.
The Security Rule takes a technology-neutral approach, specifying required and addressable implementation specifications without mandating specific technical solutions. This flexibility allows organizations to implement controls appropriate to their size, complexity, and risk profile — but it also requires documented risk analysis and risk management to justify the safeguards chosen. A HIPAA audit or enforcement action will scrutinize both your controls and your documentation of why you implemented them.
Ayliea's 80-question HIPAA assessment evaluates your implementation across the Security Rule's three safeguard categories: administrative, physical, and technical. It is designed to surface gaps before an audit or breach rather than after.
## What Ayliea Assesses
The assessment maps to the Security Rule's three safeguard categories and the major standards within each.
The largest safeguard category by question count. Covers your security management process (including risk analysis and risk management), assigned security responsibility, workforce training and access management, information access controls, security incident procedures, contingency planning, and evaluation procedures. Required specifications are assessed as binary; addressable specifications are evaluated based on your documented rationale and chosen alternatives.
Controls protecting physical access to systems that store or process ePHI. Evaluates facility access controls, workstation use policies and security, and device and media controls — including disposal, re-use, and accountability for hardware assets containing ePHI.
Access control, audit controls, integrity controls, authentication, and transmission security. Questions assess encryption at rest and in transit, unique user identification, emergency access procedures, automatic logoff, and encryption of ePHI transmitted over open networks.
Business associate agreements, group health plan requirements, and the existence of documented policies and procedures. Evaluates whether your documentation is current, accessible to workforce members who need it, and retained for the required six-year period.
## Who Needs This Assessment
* Healthcare providers (hospitals, clinics, physician practices, mental health providers) that transmit health information electronically
* Health plans and health insurance companies
* Healthcare clearinghouses that process health information
* Business associates — IT vendors, billing companies, cloud providers, legal firms, and any other entity that handles ePHI on behalf of a covered entity
* Organizations preparing for an HHS Office for Civil Rights (OCR) audit
* Healthcare organizations conducting their annual HIPAA risk analysis as required by the Security Rule
* Any organization that has recently experienced a breach and is remediating identified gaps
## Getting Started
The HIPAA Security Rule assessment requires an Organization subscription. Select **HIPAA** from the framework list in the Assess tab. The assessment is structured around the three safeguard categories, making it straightforward to assign different sections to the people in your organization with operational responsibility for each area.
For step-by-step instructions, see the [Quickstart guide](/getting-started/quickstart).
# ISO/IEC 27001:2022
Source: https://docs.ayliea.com/frameworks/iso-27001
The international standard for information security management systems (ISMS), widely required for enterprise vendor assessments and international business.
# ISO/IEC 27001:2022
ISO/IEC 27001 is published jointly by the International Organization for Standardization (ISO) and the International Electrotechnical Commission (IEC). It defines the requirements for establishing, implementing, maintaining, and continually improving an Information Security Management System (ISMS) — a systematic approach to managing sensitive information so that it remains secure.
The 2022 revision is the current version, replacing ISO 27001:2013. It reorganized the Annex A control set from 114 controls across 14 domains to 93 controls across 4 themes (Organizational, People, Physical, and Technological), reflecting how modern security programs are structured. The revision also introduced 11 new controls covering areas such as threat intelligence, data masking, configuration management, and cloud security.
Unlike compliance frameworks that define a fixed checklist of required controls, ISO 27001 certification requires demonstrating a functioning management system — documented policies, a risk-based approach to control selection, evidence of continuous improvement, and an independent audit. Ayliea's 110-question assessment evaluates both your Annex A control implementation and the governance structures that underpin a certifiable ISMS.
## What Ayliea Assesses
The assessment covers the ISMS governance requirements alongside the four Annex A control themes.
The management system requirements that must be in place for any ISO 27001 certification. Evaluates context of the organization, leadership commitment and policy, planning (risk assessment and treatment), support (resources, awareness, communication, documented information), operational controls, performance evaluation (internal audit, management review), and continual improvement processes.
37 controls covering policies, roles and responsibilities, threat intelligence, information security in supplier relationships, incident management, business continuity, and legal and regulatory compliance. Includes the 11 new controls introduced in 2022.
8 controls governing the human factor: screening, terms of employment, security awareness training, disciplinary processes, responsibilities after termination, and remote working security.
14 controls protecting physical environments — physical security perimeters, entry controls, securing offices and facilities, protection against physical and environmental threats, equipment maintenance, and secure disposal.
34 controls covering endpoint security, privileged access management, access control, cryptography, secure development, vulnerability management, network security, and monitoring. Includes new controls for configuration management, data masking, and web filtering.
## Who Needs This Assessment
* Organizations pursuing ISO 27001 certification for the first time — use the assessment to identify gaps before engaging a certification body
* Currently certified organizations conducting internal audits between surveillance audits
* Procurement and vendor management teams evaluating supplier ISMS maturity
* SaaS companies, cloud providers, and IT services firms that face ISO 27001 requirements from enterprise customers
* Organizations operating internationally or in markets where ISO 27001 is a standard vendor qualification requirement
* Security teams that want a management-system-oriented complement to control-focused frameworks like CIS or NIST
## Getting Started
The ISO 27001:2022 assessment requires an Organization subscription. Select **ISO 27001** from the framework list in the Assess tab. The ISMS governance section is best completed by whoever owns your information security program; the Annex A control themes can be distributed to technical leads responsible for each area.
For step-by-step instructions, see the [Quickstart guide](/getting-started/quickstart).
# ISO/IEC 42001
Source: https://docs.ayliea.com/frameworks/iso-42001
The international AI management system standard — governance, risk, lifecycle, and operations for organizations deploying AI responsibly.
# ISO/IEC 42001 — AI Management System
ISO/IEC 42001:2023 is the world's first international standard for an **AI Management System (AIMS)**. It applies the same management-system pattern used by ISO 27001 (information security) and ISO 9001 (quality) to AI: a documented, audited, continuously-improving system that governs how an organization designs, deploys, and operates AI.
ISO 42001 is the natural certification path for organizations that need a formal, third-party-auditable AI governance program — regulated industries, EU AI Act high-risk system providers, and enterprises whose customers require AI-program attestation.
ISO 42001 certification requires a registered certification body audit. Ayliea's assessment maps your current state to the standard's clauses + Annex A controls, identifies gaps, and produces evidence you can carry into the certification audit.
## What this framework covers
The assessment is organized around the standard's six management-system clauses plus Annex A controls. Each clause maps to one or more sub-clauses with structured questions.
Understanding the organization and its context, interested parties, AIMS scope, top-management leadership, the AI policy, and assigned roles + responsibilities + authorities.
Risk and opportunity treatment, AI risk assessment processes, AI system impact assessment, AI objectives, and planning of AIMS changes.
Resources, competence, awareness, communication, and documented information requirements for the AIMS.
Operational planning and control, AI risk assessment and impact assessment in execution.
Monitoring, measurement, analysis, evaluation, internal audit, and management review of the AIMS.
Nonconformity, corrective action, and continual improvement of the AIMS.
The controls catalog covering AI policy, internal organization, resources, impact assessment, lifecycle, data, third-party, customer information, and use of AI systems.
## Why this matters for customers
ISO 42001 gives an organization a single, auditable framework to demonstrate "we govern our AI responsibly" — backed by the same management-system rigor that regulated industries already trust for information security (ISO 27001). For customers selling into the EU, financial services, healthcare, or government markets, an ISO 42001 program shortens the procurement-review cycle on every deal.
This assessment surfaces:
* Whether your AI policy, scope, and risk approach are documented at the level a certification audit expects
* Whether AI impact assessments are conducted before high-risk systems go live
* Whether your AIMS has the leadership commitment, resources, and review cadence the standard requires
* Whether your operational controls (Annex A) are implemented and evidenced
## How it relates to other frameworks
ISO 42001 is the management-system layer. For the technical control specifics, pair it with:
* **AI Security (AISS)** — control-level technical depth on AI security
* **NIST AI RMF** — the risk-management lifecycle that ISO 42001's clauses formalize
* **NIST AI 600-1 (GAI Profile)** — generative-AI-specific risk patterns
* **ISO/IEC 27001:2022** — information security management system (sister standard for InfoSec scope)
## Glass-Box scoring
Every score cites the specific ISO 42001 clause and Annex A control behind each question. Auditors and certification bodies can map directly from your answers to the standard text.
# NIST SP 800-53 Rev 5
Source: https://docs.ayliea.com/frameworks/nist-800-53
The most comprehensive federal security control catalog, required for US federal agencies and widely adopted by federal contractors and regulated industries.
# NIST SP 800-53 Rev 5
NIST Special Publication 800-53 Revision 5 is published by the National Institute of Standards and Technology and serves as the definitive security and privacy control catalog for US federal information systems. It is the broadest and most comprehensive framework supported by Ayliea, covering 197 questions across 20 control families.
While 800-53 is mandatory for federal agencies under FISMA and FedRAMP, it has become a de facto reference framework across industries that require rigorous control documentation — including defense contractors, financial institutions, healthcare systems, and cloud service providers pursuing federal contracts. Its depth makes it the right choice for organizations that need to demonstrate exhaustive control coverage rather than just a baseline.
Revision 5 is the current release. Key changes from Rev 4 include tighter privacy control integration, the addition of supply chain risk management as a dedicated control family, and updates to reflect cloud, IoT, and mobile threat landscapes.
## What Ayliea Assesses
The assessment covers all 20 NIST 800-53 control families, evaluating both the existence of controls and the operational evidence that supports them.
Policies and technical controls governing who can access systems, under what conditions, and how identities are verified. Includes account lifecycle management, least privilege enforcement, multi-factor authentication, and session management.
Security awareness and role-based training programs, as well as audit logging — what events are captured, how logs are protected, how long they are retained, and whether they are reviewed.
Configuration baseline management and change control processes, business continuity and disaster recovery planning and testing, and your risk assessment methodology including system categorization and risk acceptance processes.
Incident response plan maturity, team structure, communication procedures, and post-incident review practices, alongside controlled maintenance of information systems and maintenance personnel access controls.
Media handling and sanitization controls, physical and environmental protection of facilities and systems, and personnel security covering background screening, access agreements, and termination procedures.
Security planning documentation, enterprise-wide program management, system security plans, and privacy risk assessments. Evaluates governance structures and executive accountability for security.
Network architecture controls — boundary protection, encryption in transit, network segmentation, and protection of communications at the application and transport layers.
Malware protection, flaw remediation, information input validation, and security alerts and advisories, as well as supply chain risk management including vendor assessments and software provenance controls.
## Who Needs This Assessment
* US federal agencies and departments required to comply with FISMA
* Organizations pursuing FedRAMP authorization to sell cloud services to the federal government
* Defense contractors subject to CMMC or DFARS requirements that reference NIST 800-53
* Cloud service providers and SaaS companies with federal agency customers
* Healthcare and financial institutions that use 800-53 as their primary control catalog
* Organizations conducting gap assessments before a formal government security authorization
## Getting Started
The NIST SP 800-53 assessment requires an Organization subscription. Select **NIST SP 800-53** from the framework list in the Assess tab. Given its scope of 197 questions, most organizations complete it across multiple sessions — answers save automatically and you can resume at any point.
For step-by-step instructions, see the [Quickstart guide](/getting-started/quickstart).
# NIST AI RMF
Source: https://docs.ayliea.com/frameworks/nist-ai-rmf
NIST's foundational risk-management framework for AI — Govern, Map, Measure, Manage — adopted across U.S. federal AI programs and increasingly across industry.
# NIST AI Risk Management Framework
The NIST AI Risk Management Framework (AI RMF 1.0) is the U.S. government's foundational framework for managing AI risk. Organized around four core functions — **Govern, Map, Measure, Manage** — it gives organizations a structured lifecycle for identifying, assessing, and treating AI risks across systems and use cases.
NIST AI RMF is the right framework for organizations adopting AI broadly, especially those that:
* Sell into U.S. federal markets or work with federal contractors
* Need a vendor-neutral, well-cited framework recognized by auditors and regulators
* Want a lifecycle structure that complements ISO/IEC 42001 certification work
Many AI-specific regulations (EU AI Act guidance, Colorado AI Act, state-level frameworks) explicitly reference NIST AI RMF as an acceptable risk-management approach. A strong AI RMF posture often translates directly to other regulatory contexts.
## What this framework covers
The assessment spans the four core functions, with categories grouped by the lifecycle phase they govern.
Organizational policies, risk management processes, and oversight structures for AI systems. Covers legal compliance, resource allocation, documentation, and system inventory.
Workforce diversity, stakeholder engagement, organizational risk culture, and third-party AI risk management practices.
Establishing context for AI system risks — categorizing systems, identifying potential harms, capturing benefits and costs, and setting risk tolerance for each system class.
Selecting appropriate metrics and methods to measure identified AI risks — including effectiveness of risk controls, system trustworthiness characteristics, and quality of inputs and outputs.
Treating risks based on measurement, prioritizing remediation, communicating risk to stakeholders, and continuously improving the AI risk management posture.
## Why this matters for customers
NIST AI RMF gives an organization a defensible answer to "how do you manage AI risk?" The framework is broad enough to apply to a wide range of AI use cases (predictive, generative, agentic) while being prescriptive enough to drive concrete program changes.
This assessment surfaces:
* Whether your AI governance policies cover the full lifecycle from sourcing through retirement
* Whether your AI system inventory captures categorization and risk classification
* Whether you measure risk with metrics appropriate to each system class (not one-size-fits-all)
* Whether risk treatment decisions are tracked, communicated, and revisited as systems evolve
## How it relates to other frameworks
NIST AI RMF is the foundational lifecycle framework. Pair it with these specializations for depth:
* **NIST AI 600-1 (GAI Profile)** — generative-AI-specific risks layered on the AI RMF core
* **ISO/IEC 42001** — the formal AI management system standard (certification-ready)
* **AI Security (AISS)** — control-level technical depth for AI security specifically
* **AI Agent Security** — agent-specific guardrails for the LLM06 Excessive Agency risk class
## Glass-Box scoring
Each question cites the specific NIST AI RMF function and category, with the published action catalog used as the source of truth. Auditors can map directly from your answers to the NIST text.
# NIST AI 600-1 — Generative AI Profile
Source: https://docs.ayliea.com/frameworks/nist-ai-rmf-gai-profile
NIST's Generative AI Profile addressing 12 GAI-specific risks layered on the AI RMF core — confabulation, CBRN risks, data privacy, value chain, and more.
# NIST AI 600-1 — Generative AI Profile
The NIST Generative AI Profile (NIST AI 600-1, July 2024) extends the AI RMF core with specific guidance for generative AI systems. While the foundational AI RMF applies broadly across AI categories, the GAI Profile dives into the 12 risks that are unique to or amplified by generative AI — confabulation, CBRN information access, value-chain integrity, data privacy, and more.
The GAI Profile is the right framework when your organization deploys generative AI (chatbots, copilots, content generation, code generation, document analysis) and needs a structured way to address risks specific to that class of system.
The GAI Profile complements rather than replaces the AI RMF. Most organizations using generative AI should run both: AI RMF for the lifecycle structure, AI 600-1 for the GAI-specific risk patterns.
## What this framework covers
The assessment is organized around the 12 GAI-specific risks defined in the NIST publication, with actions drawn from the NIST AI 600-1 action catalog mapped to AI RMF Govern, Map, Measure, and Manage functions.
Eased access to or synthesis of materially nefarious information or design capabilities related to chemical, biological, radiological, or nuclear (CBRN) weapons or other dangerous materials or agents.
The production of confidently stated but erroneous or false content by which users may be misled or deceived.
GAI systems producing dangerous, violent, hateful, or otherwise harmful content.
Generative AI systems handling personal data, sensitive data, or information that could be used to re-identify individuals.
Energy consumption, water consumption, and broader environmental impacts of training and operating GAI systems.
Bias in GAI outputs, homogenization of outputs across similar prompts, and the disparate impact of those patterns.
The configurations of human roles and oversight around GAI systems, including reliance on outputs and erosion of human judgment.
Synthetic content provenance, deepfakes, and the integrity of information ecosystems GAI systems contribute to.
Security risks specific to GAI: prompt injection, model theft, evasion of safety measures, and supply-chain compromise of foundation models.
Copyright, trademark, and trade-secret considerations in GAI training data and outputs.
GAI systems used to produce content that is obscene, degrading, abusive, or otherwise socially harmful.
Risks introduced through the GAI value chain — foundation model providers, fine-tuning vendors, deployment platforms, and downstream integrators.
## Why this matters for customers
The GAI Profile gives a structured way to answer "what could go wrong with our generative AI deployment, and what controls are we using to mitigate each risk?" — at a level of specificity that broader frameworks don't reach.
This assessment surfaces:
* Whether your data-privacy controls account for prompts and retrieved context, not just training data
* Whether your safety testing covers confabulation, jailbreak resistance, and disallowed-content production
* Whether your value-chain assessment captures foundation-model provider risks
* Whether human oversight in your GAI workflows reflects realistic reliance patterns
## How it relates to other frameworks
The GAI Profile is the generative-AI specialization layered on NIST AI RMF. Pair it with:
* **NIST AI RMF** — the foundational lifecycle framework (run both for comprehensive AI risk coverage)
* **AI Security (AISS)** — control-level technical depth, including AC-3.7 (RAG/vector stores), AC-6.6 (provenance/deepfake disclosure), AC-10.6 (agentic guardrails)
* **OWASP LLM Top 10** — the community-standard catalog of LLM application security risks
## Glass-Box scoring
Each question cites the specific GAI Profile risk category and the underlying NIST AI 600-1 action ID. The full action catalog (211 action IDs) is reflected in the scoring.
# NIST Cybersecurity Framework 2.0
Source: https://docs.ayliea.com/frameworks/nist-csf
A risk-based framework published by NIST that organizes security capabilities into six core functions, suitable for organizations of any size or industry.
# NIST Cybersecurity Framework 2.0
The NIST Cybersecurity Framework (CSF) was originally published in 2014 in response to a Presidential Executive Order and has since become one of the most widely adopted security frameworks globally. Version 2.0, released in 2024, expands the framework beyond critical infrastructure to be applicable to any organization regardless of size, sector, or maturity level.
Unlike NIST SP 800-53, which is a prescriptive control catalog, the CSF is a risk-based framework designed to help organizations understand, assess, and communicate their cybersecurity posture. It provides a common language for security conversations between technical teams and leadership, making it particularly valuable for organizations that need to communicate risk at the board or executive level.
CSF 2.0's most significant update is the addition of a sixth core function: **Govern**. This addition elevates cybersecurity governance — roles, policies, risk management strategy, and executive accountability — to the same level as the operational functions. Ayliea's 119-question assessment covers all six functions.
## What Ayliea Assesses
The assessment is organized around the six CSF core functions, each representing a distinct set of cybersecurity outcomes.
New in CSF 2.0. Evaluates whether your organization has the governance structures, policies, roles, and risk management strategy in place to guide cybersecurity decisions. Includes board-level accountability, supply chain risk governance, and cybersecurity strategy alignment with business objectives.
How well your organization understands its assets, risks, and threat environment. Questions assess asset inventory completeness, business environment mapping, risk assessment processes, and improvement planning.
The controls your organization has in place to limit the impact of a cybersecurity event. Covers identity and access management, awareness training, data security, platform security, and resilience of technology infrastructure.
Your ability to identify cybersecurity events in a timely manner. Evaluates continuous monitoring capabilities, adverse event detection, and monitoring of your supply chain and business partners.
What happens when an incident occurs. Assesses incident response plan coverage, communication procedures, analysis capabilities, mitigation processes, and post-incident improvement practices.
Your ability to restore normal operations after a cybersecurity incident. Questions evaluate recovery plan completeness, backup and restoration capabilities, communication during recovery, and lessons-learned integration.
## Who Needs This Assessment
* Organizations wanting a risk-based security posture assessment that maps to business objectives
* Security and GRC teams that need a common language for communicating risk to executives and boards
* Organizations at any maturity level — CSF scales from early-stage programs to mature security operations
* Companies with customers, partners, or regulators that reference CSF in their requirements
* Organizations using CSF as a complement to more prescriptive frameworks like NIST 800-53 or CIS Controls
* Any organization that wants a framework-neutral starting point before committing to a specific compliance standard
## Getting Started
The NIST CSF assessment requires an Organization subscription. Select **NIST CSF 2.0** from the framework list in the Assess tab. The six-function structure makes it natural to complete one function per session.
For step-by-step instructions, see the [Quickstart guide](/getting-started/quickstart).
# Compliance Frameworks
Source: https://docs.ayliea.com/frameworks/overview
Assess your security and AI governance posture across 11 active frameworks — from CIS Controls and NIST 800-53 to NIST AI RMF and OWASP LLM Top 10.
# Compliance Frameworks
Ayliea supports 11 active frameworks spanning traditional cybersecurity governance, AI-specific risk management, and emerging generative-AI standards. Each framework includes structured assessment questions, weighted scoring, and AI-powered remediation recommendations.
## Available now
The open Ayliea AI Security Standard — 10 control domains, 59 sub-controls, published under CC-BY-4.0 and crosswalked to 6 reference frameworks.
Controls for AI agents acting with delegated authority — least-privilege tool grants, blast-radius limits, prompt-injection defenses, and audit trails.
Enterprise security best practices across 18 control areas, mapping to widely-cited implementation tiers.
The comprehensive federal security control catalog, Moderate baseline.
Organized by the Govern, Identify, Protect, Detect, Respond, and Recover functions.
NIST's foundational risk-management framework for AI: Govern, Map, Measure, Manage.
NIST's Generative AI Profile addressing 12 GAI-specific risks layered on the AI RMF core.
The international AI management system standard — governance, risk, lifecycle, and operations for AIMS.
The community-standard LLM application security risk catalog.
Security Rule controls for healthcare organizations handling electronic protected health information.
Trust Services Criteria for service organizations: security, availability, processing integrity, confidentiality, privacy.
## Coming soon
Information security management system requirements. Coverage and crosswalks built; final scoring tuning in progress.
Payment card data protection. Coverage and crosswalks built; final scoring tuning in progress.
The healthcare-sector common security framework. On the roadmap — not yet available to assess against.
## How assessments work
1. **Select a framework** — choose the standard relevant to your environment, whether that's a traditional security framework, an AI-specific one, or a combination.
2. **Answer questions** — each question maps to a specific control or sub-control with weighted scoring. Most questions are multiple-choice with maturity tiers; some are yes/no with evidence prompts.
3. **Review your score** — see your overall posture grade (A through F) with category breakdown, weak spots highlighted, and per-control details surfaced.
4. **Follow recommendations** — AI-powered remediation guidance prioritized by impact, with cited frameworks and clear next steps.
5. **Track progress** — reassess over time to measure improvement and capture an audit trail.
## Cross-framework mapping
Controls overlap across frameworks. Ayliea maps these relationships so a single assessment can estimate your readiness across multiple standards. The AISS framework crosswalks to NIST CSF 2.0, NIST AI RMF, CIS Controls v8.1, ISO/IEC 27001:2022, the EU AI Act, and the Colorado AI Act — completing one AISS assessment surfaces partial coverage across all six. AISS domains are also mapped at the technique level to MITRE ATLAS for threat-informed context.
The Glass-Box drilldown on every results page shows the exact framework references behind each question, so an auditor can verify any score against the cited source.
## Free vs. engagement access
The **AI Security (AISS)** framework is available for free — any organization can run it without payment or an engagement. Every other framework is available once your organization has an active engagement; which frameworks are in scope depends on what you and your assessor agreed to. Contact your Ayliea engagement lead if you're unsure what's included.
# OWASP LLM Top 10
Source: https://docs.ayliea.com/frameworks/owasp-llm-top-10
The community-standard catalog of LLM application security risks — prompt injection, sensitive information disclosure, training data poisoning, and more.
# OWASP LLM Top 10
The OWASP Top 10 for LLM Applications is the community-standard catalog of the most critical security risks in applications that integrate large language models. Maintained by the OWASP Foundation with input from practitioners and researchers across the LLM security community, it's the de-facto reference for "what could go wrong with an LLM-powered feature."
This assessment is appropriate for any organization shipping LLM-powered features — customer-facing chatbots, internal copilots, RAG-based search and Q\&A, code-assistance tools, document analyzers, or any application where user input flows through an LLM.
OWASP LLM Top 10 v2025 (published 2025-03) is the current version. The framework is maintained on a regular cadence — Ayliea tracks updates and refreshes the assessment as the catalog evolves.
## What this framework covers
The assessment maps directly to the 10 risks in the OWASP catalog. Each risk has its own control area with questions tailored to that risk's specific attack patterns and mitigations.
User-supplied content alters the LLM's intended behavior — direct injection, jailbreaks, indirect injection via retrieved or tool-output content, and output manipulation that subverts downstream actions.
PII, PHI, credentials, proprietary data, or business-confidential context surfaced through LLM outputs — whether memorized from training, leaked via RAG retrieval, or echoed back from the user's own session.
Compromises in the LLM value chain: foundation model providers, fine-tuning datasets, embeddings models, tool integrations, and the deployment infrastructure connecting them.
Adversarial manipulation of training data, fine-tuning data, or retrieved context to bias outputs, embed backdoors, or degrade model behavior.
Downstream systems trusting LLM output without validation — leading to XSS, SQL injection, command injection, server-side request forgery, or similar classical web-app risks rendered via LLM intermediation.
LLM-powered agents granted more permissions than their use case requires — leading to consequential actions taken without sufficient human oversight, blast-radius limits, or revocation paths.
System prompts and tool definitions leaked to users, exposing implementation details, business logic, or guardrail boundaries that attackers can use to refine prompt-injection attempts.
Weaknesses in RAG and vector-store deployments: poisoning the corpus, embedding inversion attacks, cross-tenant retrieval leaks, and stale or untrusted retrieved content treated as authoritative.
LLM outputs presented as authoritative when they're confabulated, outdated, or systematically biased — including reliance failures where users defer to LLM judgment in high-stakes contexts.
Denial-of-wallet attacks via excessive inference requests, expensive tool invocations, or unbounded recursion patterns that aren't rate-limited or cost-bounded server-side.
## Why this matters for customers
Engineering teams shipping LLM features need a shared reference for "what we tested for" — both for internal pre-launch review and for external customer security questionnaires. The OWASP LLM Top 10 has become that reference. A clean assessment is increasingly table stakes for any vendor selling LLM-powered features into security-conscious buyers.
This assessment surfaces:
* Whether your prompt-injection defenses cover direct, indirect, and tool-output injection paths
* Whether your RAG deployment isolates tenants and refreshes the corpus on a defined cadence
* Whether your output handling validates LLM responses before downstream systems trust them
* Whether your cost and rate controls protect against denial-of-wallet patterns
## How it relates to other frameworks
OWASP LLM Top 10 is the practitioner-focused application security catalog. Pair it with:
* **AI Security (AISS)** — broader AI program controls including governance, asset management, and incident response
* **AI Agent Security** — drills deeper into LLM06 Excessive Agency for agent-specific deployments
* **NIST AI 600-1 (GAI Profile)** — the regulatory-aligned risk-management view of many of the same risks
* **NIST AI RMF** — the foundational AI risk-management lifecycle
## Glass-Box scoring
Each question cites the specific OWASP LLM Top 10 risk ID (LLM01–LLM10) and version (currently 2025-03). Where the risk pattern maps to MITRE ATLAS techniques, those are surfaced in the drilldown as well.
# PCI DSS v4.0
Source: https://docs.ayliea.com/frameworks/pci-dss
Assessment against the Payment Card Industry Data Security Standard, required for any organization that processes, stores, or transmits payment card data.
# PCI DSS v4.0
The Payment Card Industry Data Security Standard (PCI DSS) is published and maintained by the PCI Security Standards Council, a body founded by the major card brands (Visa, Mastercard, American Express, Discover, and JCB). It defines the security requirements that any organization must meet if it processes, stores, or transmits payment card data — regardless of company size, transaction volume, or industry.
Version 4.0 is the current release, representing the most significant revision to PCI DSS in over a decade. It introduces a customized approach that allows mature organizations to achieve requirements through alternative controls with documented validation, adds new requirements targeting phishing and e-commerce security, and expands multi-factor authentication requirements across the cardholder data environment. Compliance with PCI DSS v4.0 became mandatory in March 2024.
Non-compliance carries significant consequences: fines from acquiring banks, increased transaction fees, mandatory forensic investigations following a breach, and potential loss of card acceptance privileges. Ayliea's 124-question assessment covers all 12 PCI DSS requirements across the six control objectives, helping you identify gaps before a formal Qualified Security Assessor (QSA) engagement.
## What Ayliea Assesses
The assessment maps to the six PCI DSS control objectives and 12 requirements within them.
Network security controls protecting the cardholder data environment (CDE). Evaluates firewall configuration, network segmentation, documentation of allowed traffic flows, and whether default vendor passwords and security settings have been changed on all system components.
How your organization stores and transmits cardholder data. Covers data retention and disposal policies, encryption of stored primary account numbers (PAN), masking on displays, and encryption of cardholder data transmitted over open, public networks. Includes key management practices.
Malware protection across system components, vulnerability management processes, and secure software development practices. Evaluates scanning frequency, patch management SLAs, and software security controls including protection against common web application vulnerabilities.
Restricting access to cardholder data on a need-to-know basis. Covers logical access controls, unique user IDs and authentication requirements (including the expanded MFA requirements in v4.0), and physical access controls for facilities and systems in the cardholder data environment.
Logging and monitoring of all access to network resources and cardholder data, audit log protection and review procedures, and regular testing of security systems and processes — including internal and external vulnerability scanning, penetration testing, and intrusion detection coverage.
The policy and governance foundation of your PCI DSS program. Evaluates your information security policy coverage and annual review process, risk assessments, security awareness training, incident response plan, and third-party service provider management including contractual requirements and annual acknowledgment of responsibility.
## Who Needs This Assessment
* E-commerce companies that accept card payments on their websites or applications
* Retail and hospitality organizations with point-of-sale systems
* Payment processors, payment gateways, and acquiring banks
* SaaS companies that integrate with payment systems on behalf of their customers
* Organizations that store card data for recurring billing or subscription management
* Companies preparing for an annual QSA assessment or Self-Assessment Questionnaire (SAQ)
* Organizations that have experienced a payment card breach and are implementing remediation requirements
PCI DSS applicability and the level of assessment required (QSA audit vs. SAQ) depends on your annual transaction volume and card brand agreements. Consult your acquiring bank or a QSA to determine which compliance path applies to your organization. Ayliea's assessment is a gap analysis tool and does not replace a formal QSA engagement.
## Getting Started
The PCI DSS v4.0 assessment requires an Organization subscription. Select **PCI DSS** from the framework list in the Assess tab. The six control objective structure makes it natural to assign sections to your network, application security, and GRC teams.
For step-by-step instructions, see the [Quickstart guide](/getting-started/quickstart).
# SOC 2 Type II Security TSC
Source: https://docs.ayliea.com/frameworks/soc-2
Assessment against the AICPA Trust Service Criteria for security, the baseline requirement for SaaS companies and cloud providers serving enterprise customers.
# SOC 2 Type II Security TSC
SOC 2 (System and Organization Controls 2) is a reporting framework developed by the American Institute of Certified Public Accountants (AICPA). It evaluates service organizations against Trust Service Criteria (TSC) — a set of principles covering Security, Availability, Processing Integrity, Confidentiality, and Privacy. The Security TSC (also called the Common Criteria) is the baseline required in every SOC 2 engagement and is what Ayliea's 84-question assessment evaluates.
A SOC 2 Type II report covers a period of time (typically 6 to 12 months) rather than a point in time. The report provides evidence that your controls were not just designed correctly but operated effectively throughout the review period. This distinction makes SOC 2 Type II the gold standard that enterprise buyers, investors, and partners request when assessing the security of service organizations.
If your organization is a SaaS company, cloud platform, data processor, or any business that stores or processes customer data, you will almost certainly be asked to provide a SOC 2 report at some point in a sales cycle or enterprise vendor assessment. Ayliea's assessment helps you understand your readiness before you engage an auditor.
## What Ayliea Assesses
The assessment maps to the AICPA's Common Criteria (CC) organized into seven logical groupings.
The foundation of your internal control program. Evaluates board and management commitment to security, assignment of authority and responsibility, human resources practices, and the overall tone and accountability structures that support effective controls.
How your organization obtains, generates, and uses information relevant to security objectives. Covers internal communication of security policies and responsibilities and external communication of commitments to customers and third parties.
Your process for identifying and assessing risks to achieving security commitments. Evaluates risk identification methodology, fraud risk consideration, and how risk assessment informs control design and investment.
How your organization evaluates whether controls are present and functioning. Covers ongoing monitoring activities, independent evaluations, and how deficiencies are identified, reported, and remediated.
The selection and development of specific controls that mitigate identified risks. Evaluates how controls are designed, implemented, and updated in response to changing risk conditions.
The most technical section of the Common Criteria. Covers logical and physical access restrictions to assets, encryption and key management, network and endpoint security, change management processes (including software development lifecycle controls), and risk mitigation for vendor and business partner access.
Business disruption risk identification and mitigation, and vendor and business partner risk management. Evaluates how your organization monitors for threats, responds to incidents, and manages the risks introduced by the third parties you depend on.
## Who Needs This Assessment
* SaaS companies that sell to enterprise customers who require SOC 2 reports as a condition of purchase
* Cloud service providers, data centers, and managed security service providers
* Any organization that stores, processes, or transmits sensitive customer data
* Companies preparing for a SOC 2 Type I or Type II audit — use the assessment to identify gaps before the auditor arrives
* Organizations responding to security questionnaires that reference SOC 2 criteria
* Compliance teams that need to map existing controls to the Common Criteria before a readiness assessment
## Getting Started
The SOC 2 assessment requires an Organization subscription. Select **SOC 2** from the framework list in the Assess tab. The logical groupings make it straightforward to assign the assessment across your engineering, security, and GRC teams.
For step-by-step instructions, see the [Quickstart guide](/getting-started/quickstart).
# Getting Access
Source: https://docs.ayliea.com/getting-started/creating-account
How access to Ayliea is provisioned for engagement clients, and how to activate your account
Ayliea doesn't have public self-serve sign-up. Access is provisioned as part of starting an engagement — your Ayliea contact sets up your organization and your account after your engagement is scoped, and you activate your own account from there.
## How access works
1. **You start an engagement.** After a scoping call, your organization is set up in Ayliea and an Owner account is created for you (or the person your team designates).
2. **You receive an activation email.** The email contains a secure, time-limited activation link addressed to your work email.
3. **You complete activation.** Following the link walks you through a short guided setup — no separate sign-up form to fill out.
If you've been invited to join an existing organization as a teammate rather than as the first Owner, the flow is the same: you'll receive an activation email and follow the same guided setup.
## Activating your account
Click the activation link. It's valid for a limited time — if it's expired, contact your Ayliea account team for a new one.
Choose a password. Ayliea follows NIST SP 800-63B guidelines: a **12-character minimum**, with no mandatory complexity rules — a long passphrase works as well as a mix of symbols. The setup screen shows real-time strength feedback as you type.
Enroll a TOTP authenticator app (such as Google Authenticator, 1Password, or Authy) by scanning the QR code shown.
You'll be shown a set of one-time backup codes. Store them somewhere safe — they let you sign in if you lose access to your authenticator app.
Activation is complete. Sign in with your email, password, and a code from your authenticator app.
Multi-factor authentication is part of account setup for every Ayliea account. See [ayliea.com/trust](https://ayliea.com/trust) for Ayliea's full security posture, including password policy and encryption details.
## What you get access to
What frameworks, portal features, and governance modules are available to your organization depends on your engagement scope. See [Frameworks Overview](/frameworks/overview) for what's available without an engagement versus with one, and [Organizations](/product/organizations) for how your workspace and team access are structured. If you're unsure what's included in your engagement, contact your Ayliea engagement lead.
# Quickstart
Source: https://docs.ayliea.com/getting-started/quickstart
Get started with Ayliea's client portal after your organization is provisioned
This guide covers the client portal used during an agreed engagement. If you are exploring Ayliea before an engagement, the [free AI readiness quiz](https://ayliea.com/assess) gives a self-reported snapshot without an account. It is separate from the portal and from an independent assessment report.
Start with a [scoping conversation](https://ayliea.com/contact). When portal access is part of the engagement, Ayliea provisions your organization and sends an activation invitation. See [Getting Access](/getting-started/creating-account).
Sign in and use the **Assess** tab to open the framework or assessment included in your engagement. Available content depends on your scope. See the [Frameworks Overview](/frameworks/overview) for reference.
Work through the questions and record relevant context. Your engagement lead will explain which evidence to provide and the approved transfer method. Do not send sensitive assessment material through the public contact form.
Portal scores and recommendations help organize the work. The assessor's delivered report identifies the evidence reviewed, findings, limitations, and prioritized remediation within the agreed scope. See [Your First Assessment](/getting-started/your-first-assessment) for the portal walkthrough.
# Your First Assessment
Source: https://docs.ayliea.com/getting-started/your-first-assessment
Walk through an assigned assessment in the Ayliea client portal
## Selecting a Framework
This walkthrough applies to an organization with provisioned portal access. Open the **Assess** tab to view the assessment content available under your engagement. Each framework card shows its name, a brief description, and the number of categories and questions it covers.
Select the assessment agreed with your Ayliea engagement lead. The [Frameworks Overview](/frameworks/overview) explains the available references, but a framework's presence in the documentation does not mean it is included in every engagement.
If you do not have portal access, you can use the [free AI readiness quiz](https://ayliea.com/assess) without an account. Its self-reported result is separate from an assessor's evidence-based findings and report.
## Answering Questions
Tap any category to begin. Questions are presented one at a time with a progress bar and question navigator at the top. The navigator shows which questions you've answered and lets you jump to any item directly.
You can resume an in-progress assessment at any time. Answers save automatically as you select them — you will never lose progress if you close the app mid-category.
Ayliea uses four question types depending on the control being evaluated:
A binary compliance check. Select **Yes** if the control is fully in place or **No** if it is not. This type is used for clear-cut requirements where partial credit does not apply.
Select the one option that best describes your current state. Each option carries a different point value — more mature or complete implementations score higher.
Select all sub-controls or practices that apply to your environment. Partial credit is awarded based on the number of items checked, so even incomplete implementations contribute to your score.
Rate your maturity on a 1–5 scale. 1 indicates no controls in place; 5 indicates a fully documented, tested, and continuously monitored practice.
Every question also has an optional **Notes** field where you can record evidence references, ticket numbers, or caveats for your records.
## Completing a Category
Tap **Next** after each answer. When you reach the last question in a category, the button changes to **Finish Category** and returns you to the category list. Completed categories display a checkmark.
Repeat this process for each category in the framework. You do not need to finish all categories in one session.
## Viewing Your Results
Once all categories in a framework are marked complete, Ayliea calculates your score and navigates to the Results screen. You will see:
* **Score gauge** — your overall weighted score from 0–100 with a letter grade (A–F)
* **Category breakdown** — a bar chart of each category's score, sorted weakest-first to prioritize attention
* **AI recommendations** — a ranked list of remediation actions generated from your weakest control areas, with implementation playbooks where available
Tap any recommendation to view its full detail, including effort estimate, business impact, and step-by-step guidance.
# Incident Tracking
Source: https://docs.ayliea.com/governance/incidents
Report, investigate, and resolve AI-related incidents
Incident Tracking is included based on your engagement scope. Contact your Ayliea engagement lead to confirm it's part of yours.
Incident tracking gives your team a structured process for managing AI-related failures, harms, and near-misses from first report to formal closure. Every incident produces a documented record with investigation findings, remediation steps, and regulatory notifications — giving you the audit trail that regulators and auditors expect.
## Reporting an incident
To report a new incident:
1. Navigate to **Governance → Incidents** and select **Report Incident**
2. Enter a title and a description of what occurred
3. Set the severity — Critical, High, Medium, or Low
4. Select a category (for example, bias/fairness, data breach, model failure, or safety violation)
5. Optionally link the incident to one or more registered AI systems
6. Select **Submit**
Once submitted, the incident is assigned an ID and enters the **Open** status.
## Incident lifecycle
Incidents move through a defined lifecycle:
| Status | Meaning |
| - | - |
| **Open** | Reported but investigation has not started |
| **Investigating** | Actively gathering information about the cause |
| **Containment** | The immediate impact is being limited while investigation continues |
| **Remediation** | Root cause identified; fixes are being implemented |
| **Resolved** | Remediation complete; incident is functionally over |
| **Closed** | Formally closed with a closing note on record |
Status transitions are recorded in the incident timeline with a timestamp and the name of the person who made the change.
## Investigation fields
During investigation, you can record:
* **Root cause** — a narrative description of what caused the incident
* **Root cause category** — a structured label (for example, training data quality, deployment error, or prompt injection)
* **Contributing factors** — additional conditions that made the incident worse or harder to detect
These fields are optional while an incident is open, but must be completed before an incident can move to Remediation.
## Remediation tracking
The Remediation section of each incident tracks the steps your team is taking to fix the underlying cause and prevent recurrence:
* **Remediation steps** — a list of actions with completion status
* **Preventive measures** — controls or process changes being put in place to reduce the likelihood of recurrence
## Regulatory notification management
For incidents that trigger regulatory reporting obligations, the notification tracker lets you record:
* Which regulators must be notified
* The notification deadline for each
* Whether the notification has been sent and when
Regulatory notification deadlines under the EU AI Act and other frameworks can be as short as 72 hours for certain incident categories. Configure notifications as soon as you determine a regulatory obligation applies.
## Timeline
Every incident has an automatic activity feed that records all status changes, field updates, and comments in chronological order. The timeline cannot be edited — it is an accurate record of exactly how the incident was handled.
Team members with access to the incident can add comments to the timeline to document decisions, share findings, or communicate status.
## Closing an incident
To close an incident, select **Close Incident** from the incident detail view. Closing requires a written closing note summarizing the outcome and any open follow-up actions. The closing note is appended to the timeline and the incident moves to **Closed** status.
Closed incidents remain visible and fully searchable. They cannot be reopened, but a new incident can reference a closed one if the same issue recurs.
## Preventive recommendations
When an incident is resolved, Ayliea generates a set of preventive recommendations based on the root cause and contributing factors recorded. These recommendations appear in the incident record and can be linked to your assessment remediation backlog so the underlying control gap gets addressed as part of your regular improvement cycle.
# AI Governance
Source: https://docs.ayliea.com/governance/overview
Manage AI risk across your organization with a unified governance module
The AI Governance module gives your organization a central place to track, classify, and manage risk across every AI system you operate or procure. Rather than managing AI risk in spreadsheets or siloed tools, you get a structured workflow that connects your AI inventory to regulatory obligations, vendor accountability, and incident response.
## Capabilities
The module is made up of five connected areas:
### AI System Registry
A catalog of every AI system your organization runs or relies on. You register each system with its purpose, use cases, and data flows. The registry is the foundation that links risk classifications, vendor assessments, and incidents to specific systems.
### Risk Classification
Classify each registered system against the EU AI Act and NIST AI RMF frameworks. The system analyzes your use cases and suggests a risk tier automatically. You can accept the suggestion or override it with a documented rationale.
### Vendor Assessments
Send structured security questionnaires to your AI vendors and track their responses. Vendors receive a secure link to complete the questionnaire directly — no account required. Responses are scored automatically and assigned a risk tier when submitted.
### Incident Tracking
Report and manage AI-related incidents from initial discovery through resolution. Each incident has a full lifecycle with investigation fields, remediation steps, and regulatory notification tracking — giving you a documented record for audits.
### Regulatory Timeline
A pre-loaded view of AI regulation deadlines drawn from the EU AI Act, Colorado AI Act, and NIST AI RMF. The system suggests which regulations apply to your organization based on your registered systems and operating jurisdictions, and tracks your compliance status per milestone.
## Availability
AI Governance module features are included based on your engagement scope. Contact your Ayliea engagement lead to confirm which of the five areas are part of your engagement.
## Accessing the module
The Governance section appears in the main navigation sidebar. Select **Governance** to open the module overview, then navigate to any of the five areas using the sub-navigation.
# AI System Registry
Source: https://docs.ayliea.com/governance/registry
Catalog and track your organization's AI deployments
The AI System Registry is included based on your engagement scope. Contact your Ayliea engagement lead to confirm it's part of yours.
The registry is a structured inventory of every AI system your organization operates or depends on — including third-party tools, internally built models, and AI features embedded in SaaS products. Keeping the registry current gives your compliance, security, and procurement teams a shared source of truth about what AI is running in your environment.
## Registering a system
To add a new system to the registry:
1. Navigate to **Governance → Registry** and select **Register System**
2. Enter the system's name, owner, and a brief description of its purpose
3. Select one or more use case categories (for example, content generation, decision support, or fraud detection)
4. Describe the data flows — what data the system receives, processes, and outputs
5. Select **Save** to add the system
Once saved, the system appears in the registry list and becomes available for risk classification, vendor assessment linking, and incident reporting.
## Viewing and editing systems
Select any system in the registry list to open its detail view. From there you can update any field, view its current risk classification, see linked vendor assessments, and review any incidents associated with the system.
Changes to system details are recorded automatically so you have an accurate history of how each system has been described over time.
## Retiring a system
When a system is decommissioned, select **Retire System** from the system detail view. Retiring a system marks it as inactive without deleting its record — past risk classifications, assessments, and incidents remain visible for audit purposes.
Retired systems are hidden from the active registry list by default. Use the **Show retired** filter to include them.
## Shadow AI and Discovery
If you use the Cloud Discovery integration, unrecognized AI services detected in your environment surface as shadow AI findings. You can convert any shadow AI finding directly into a registered system from the Discovery results view, pre-filling the registry form with the information already gathered.
Run a Discovery scan before populating the registry manually — it often surfaces AI services that teams are using without formal approval, which are the highest-risk entries to capture first.
## Registry stats on the governance dashboard
The Governance dashboard shows a summary of your registry at a glance:
* Total systems registered
* Systems awaiting risk classification
* Systems with active incidents
* Systems with vendor assessments due for renewal
# Regulatory Timeline
Source: https://docs.ayliea.com/governance/regulatory
Track AI regulation deadlines and compliance status
The Regulatory Timeline is included based on your engagement scope. Contact your Ayliea engagement lead to confirm it's part of yours.
The regulatory timeline gives your organization a consolidated view of AI regulation milestones and your progress toward each one. Rather than tracking deadlines in a separate spreadsheet, you manage compliance status, assign owners, and monitor approaching dates from within your governance workspace.
## What the timeline shows
The timeline displays a list of regulatory milestones organized by regulation and effective date. Each milestone shows:
* The regulation it belongs to
* A plain-language description of what the milestone requires
* Your current compliance status
* The owner assigned to the milestone
* The target completion date (if set)
* Days remaining until the deadline
## Pre-loaded regulations
The timeline comes pre-loaded with milestones from three frameworks:
* **EU AI Act** — phased implementation milestones from the initial prohibited systems ban through the full high-risk system requirements
* **Colorado AI Act** — developer and deployer obligations and effective dates
* **NIST AI RMF** — key adoption guidance milestones and voluntary commitments
Milestones are updated as regulations are amended or clarified. You do not need to maintain the regulation content manually.
## Auto-applicability
When you set your operating jurisdictions, Ayliea cross-references your jurisdictions and the AI systems in your registry against each regulation's applicability criteria. Regulations that are likely to apply to your organization are highlighted automatically.
Auto-applicability is a suggestion — you review and confirm which regulations apply. Confirmed regulations appear in your active timeline; dismissed regulations are hidden but remain accessible if your situation changes.
## Setting your operating jurisdictions
To configure your jurisdictions:
1. Navigate to **Governance → Regulatory Timeline** and select **Settings**
2. Add the countries and US states where your organization operates or deploys AI systems
3. Save your changes — auto-applicability runs immediately and updates the timeline
## Tracking compliance status
Each milestone has a compliance status you update as your organization works toward the requirement:
| Status | Meaning |
| - | - |
| **Not started** | No work has begun on this milestone |
| **In progress** | Active work is underway |
| **Compliant** | The requirement is fully met |
| **Non-compliant** | The deadline has passed and the requirement is not met |
Status updates are recorded in the milestone history so you have a record of when compliance was achieved.
## Assigning owners and target dates
Each milestone can be assigned to a specific team member and given a target completion date. Assigned owners can see their open milestones from their Governance dashboard view, making it easy for compliance and legal team members to track their responsibilities without navigating the full timeline.
## Countdown alerts
The Governance dashboard surfaces upcoming regulatory deadlines as countdown cards. By default, milestones due within 90 days appear as alerts. Milestones due within 30 days are highlighted to draw immediate attention.
Assign owners to all active milestones as soon as you confirm which regulations apply. Unassigned milestones have no single accountable party and are more likely to be missed as deadlines approach.
# Risk Classification
Source: https://docs.ayliea.com/governance/risk-classification
Classify AI systems against EU AI Act and NIST AI RMF frameworks
Risk Classification is included based on your engagement scope. Contact your Ayliea engagement lead to confirm what's included, including AI-assisted narrative analysis.
Risk classification assigns a formal risk tier to each system in your registry. This determines what governance requirements apply — what oversight controls, documentation, and human review processes you need to have in place. Classification runs against two complementary frameworks: the EU AI Act and the NIST AI RMF.
## Auto-classification
When you open the classification view for a registered system, Ayliea analyzes the use cases and data flows you provided during registration and suggests a risk tier automatically. The suggestion is based on the categories of use, the type of decisions the system influences, and the population it affects.
Auto-classification is a starting point, not a final determination. Review the suggestion and accept it or override it with your own assessment.
## EU AI Act tiers
The EU AI Act organizes AI systems into four tiers based on the risk they pose:
| Tier | Description |
| - | - |
| **Unacceptable risk** | Systems that pose a clear threat to fundamental rights. These are prohibited under the EU AI Act. |
| **High risk** | Systems used in critical areas such as employment, essential services, law enforcement, or education. Subject to the most extensive compliance requirements. |
| **Limited risk** | Systems with specific transparency obligations — for example, chatbots that interact with people must disclose that the user is talking to an AI. |
| **Minimal risk** | Systems that pose little or no risk. Most AI tools fall here. No mandatory compliance obligations apply. |
## NIST AI RMF assessment
Alongside the EU AI Act classification, you can record an impact and likelihood assessment aligned to the NIST AI Risk Management Framework. This gives you a two-dimensional view of risk: how severe the potential harm is, and how likely it is to occur.
The NIST assessment does not produce a pass or fail result — it produces a risk profile that informs your governance priorities.
## Accepting or overriding the auto-classification
After reviewing the suggested tier, select **Accept** to confirm it, or select **Override** to record a different tier. Overrides require a written rationale, which is saved with the classification record. This rationale is available during audits to explain why your team chose a different tier than the system suggested.
## Governance requirements
Once a system is classified, the governance requirements triggered by that classification are listed on the system detail view. For example, a High-risk system under the EU AI Act will show requirements such as:
* Conformity assessment before deployment
* Human oversight mechanism documentation
* Post-market monitoring plan
* Incident reporting obligations
These requirements are informational — they describe what your organization needs to have in place, and you track completion status directly on the system record.
## AI-assisted analysis
Where included in your engagement, you can generate a narrative risk report for any classified system. The report summarizes the classification rationale, the applicable governance requirements, and recommended next steps in plain language — useful for sharing with non-technical stakeholders or for inclusion in audit documentation.
## Framework selection
From the Classification settings, you can configure which frameworks apply to your organization. If your organization is not subject to the EU AI Act, you can disable that framework and rely solely on NIST AI RMF, or vice versa.
## Classification history
Every classification event — including the original auto-suggestion, any overrides, and changes over time — is recorded in an append-only history log on the system record. This log cannot be edited or deleted, providing an accurate audit trail of how each system's risk profile has changed.
# Vendor Assessments
Source: https://docs.ayliea.com/governance/vendor-assessments
Assess your AI vendors' security posture with structured questionnaires
Vendor Assessments are included based on your engagement scope. Contact your Ayliea engagement lead to confirm it's part of yours.
A vendor assessment is a structured security questionnaire you send to an AI vendor to understand how they manage data, protect their models, and meet compliance obligations. Completing vendor assessments gives you documented evidence of third-party due diligence — a requirement under many compliance frameworks and an expectation of enterprise procurement processes.
## Creating an assessment
To start a new vendor assessment:
1. Navigate to **Governance → Vendor Assessments** and select **New Assessment**
2. Choose a questionnaire template — templates are organized by vendor type (for example, AI platform, AI-enabled SaaS, or custom model provider)
3. Enter the vendor's name and the contact email address of the person who will complete the questionnaire
4. Optionally link the assessment to a registered system in your registry
5. Set a response deadline and select **Send**
The vendor receives an email with a secure link to their response portal. You can monitor submission status from the assessment list.
## How the vendor portal works
Vendors do not need an Ayliea account to complete a questionnaire. The secure link in their email takes them directly to a response portal where they can:
* Read each question and enter their response
* Save progress and return to finish later — answers are saved automatically
* Submit when all required questions are answered
The portal does not expose any of your organization's internal information. Vendors only see the questions and their own responses.
## Question categories
Assessment questionnaires cover five areas:
* **Data handling** — how the vendor collects, stores, processes, and deletes data; data residency and retention policies
* **Security controls** — access controls, encryption practices, vulnerability management, and penetration testing cadence
* **Privacy and compliance** — GDPR, CCPA, and other applicable regulatory compliance; privacy impact assessments
* **AI model governance** — training data sourcing, model validation, bias testing, and explainability practices
* **Contractual and operational** — SLA commitments, subprocessor disclosure, incident notification obligations, and audit rights
## Scoring and risk tier
When a vendor submits their responses, Ayliea scores the questionnaire automatically and assigns a risk tier — Low, Medium, High, or Critical — based on the responses across all categories. The scoring reflects both the completeness of responses and the strength of the controls described.
You can view the score breakdown by category to identify which areas of the vendor's posture are weakest.
## Reviewing responses
After submission, open the assessment to review the vendor's responses question by question. You can:
* **Approve** the assessment if the responses are satisfactory
* **Reject** the assessment with written notes if gaps require follow-up — the vendor is notified and can resubmit
Approval and rejection are recorded in the assessment history along with any notes you add.
## Reminder emails
If a vendor has not submitted by a configurable threshold before the deadline, Ayliea automatically sends a reminder email to the contact address on file. You can also trigger a manual reminder at any time from the assessment detail view.
Link vendor assessments to registered systems so that the system detail view shows each vendor's current assessment status and expiry date. This makes it easy to catch assessments that are approaching renewal without reviewing each vendor separately.
# AWS CloudTrail Integration
Source: https://docs.ayliea.com/integrations/aws-cloudtrail
Discover AI model usage across your AWS environment by connecting CloudTrail to Ayliea.
## Overview
The AWS CloudTrail integration automatically discovers AI model usage in your AWS environment by monitoring CloudTrail events for Amazon Bedrock API calls. This gives your security team visibility into which AI models are being used, by whom, and how frequently.
## What Data Is Collected
Ayliea reads **API call metadata only** from CloudTrail:
* **Model IDs** — which Bedrock models are invoked (e.g., `anthropic.claude-3-sonnet`)
* **User identities** — IAM users or roles making the calls
* **Timestamps** — when calls occurred
* **Event names** — the API action (e.g., `InvokeModel`)
* **Region** — which AWS region the call was made in
Ayliea **never** reads prompt content, model responses, or any payload data. Only event metadata from CloudTrail is accessed.
## Prerequisites
* An AWS account with Amazon Bedrock enabled
* Permission to create IAM users and policies
* Cloud discovery included in your engagement scope
## Setup Steps
### 1. Create an IAM Policy
In the AWS Console, navigate to **IAM > Policies > Create Policy** and paste this JSON:
```json theme={null}
{
"Version": "2012-10-17",
"Statement": [
{
"Effect": "Allow",
"Action": "cloudtrail:LookupEvents",
"Resource": "*"
}
]
}
```
Name the policy something descriptive like `AylieaCloudTrailReadOnly`.
This policy grants **read-only** access to CloudTrail events. It cannot modify any AWS resources.
### 2. Create an IAM User
1. Go to **IAM > Users > Create User**
2. Name the user (e.g., `ayliea-cloudtrail-reader`)
3. Do **not** enable console access
4. Attach the `AylieaCloudTrailReadOnly` policy you created
### 3. Generate Access Keys
1. Open the new user's detail page
2. Go to **Security credentials > Access keys > Create access key**
3. Select **Third-party service** as the use case
4. Copy the **Access Key ID** and **Secret Access Key**
The secret access key is only shown once. Store it securely until you've entered it in Ayliea.
### 4. Connect in Ayliea
1. Go to **Organization > Cloud Integrations**
2. Click **Connect** on the AWS CloudTrail card
3. Enter your Access Key ID, Secret Access Key, and Region
4. Click **Test Connection** to validate
5. Click **Connect** to save
## Polling Schedule
Once connected, Ayliea polls CloudTrail **daily at 7:00 AM UTC**. Each poll covers the time since the last successful poll. The first poll covers the previous 24 hours.
## Troubleshooting
### "Invalid credentials or insufficient permissions"
* Verify the Access Key ID and Secret Access Key are correct
* Confirm the IAM user has the `AylieaCloudTrailReadOnly` policy attached
* Check that CloudTrail is enabled in the specified region
### "Failed to connect to AWS CloudTrail"
* Verify the region is correct (e.g., `us-east-1`, not `US East`)
* Check that your AWS account doesn't have restrictive SCPs blocking CloudTrail access
### No platforms discovered after connecting
* Confirm that Amazon Bedrock is being used in the specified region
* CloudTrail events may take up to 15 minutes to appear
* The first poll covers only the last 24 hours — wait for the next daily poll
## Security Considerations
* Credentials are encrypted at rest
* The IAM policy grants **read-only** access to CloudTrail events only
* Ayliea never accesses S3 buckets, Lambda functions, or any other AWS services
* You can revoke access at any time by deleting the IAM user or disconnecting in Ayliea
# Discovery Integrations
Source: https://docs.ayliea.com/integrations/cloud-discovery
How Ayliea discovers AI usage through network traffic analysis and cloud provider integrations.
## Overview
Ayliea discovers AI platform usage across your organization through multiple methods:
| Method | Source | What It Detects |
| - | - | - |
| **[Network Collector](/integrations/network-collector)** | DNS queries + TLS SNI on your network | All AI platforms (57+ providers, 177+ domains) |
| **[AWS CloudTrail](/integrations/aws-cloudtrail)** | CloudTrail LookupEvents API | Amazon Bedrock (Claude, Titan, Llama, etc.) |
| **[GCP Cloud Logging](/integrations/gcp-cloud-logging)** | Cloud Logging entries.list API | Vertex AI (Gemini, PaLM, etc.) |
The Network Collector provides the broadest coverage — it detects any AI platform accessed from your network, regardless of cloud provider. Cloud integrations provide deeper detail (specific model IDs, caller identities) for services running in your cloud accounts.
## How It Works
### Connection
1. You create a read-only service account in your cloud provider
2. Enter the credentials in Ayliea's Organization settings
3. Ayliea validates the credentials and saves them encrypted
### Daily Polling
Ayliea polls your cloud provider **once per day at 7:00 AM UTC**:
1. Reads API audit logs since the last successful poll
2. Extracts model identifiers from each API call
3. Maps model IDs to Ayliea's AI tool catalog
4. Creates a discovery scan record with the results
5. Generates alerts for newly discovered platforms
6. Checks results against your AI usage policies
### Model Mapping
Cloud providers use internal model identifiers (e.g., `anthropic.claude-3-sonnet-20240229-v1:0`). Ayliea normalizes these identifiers and maps them to entries in the AI tool catalog, which provides:
* Human-readable names
* Vendor information
* Risk classifications
* Category assignments
Unmapped models appear with their raw cloud identifier and a default risk level.
## Integration with Policy Engine
Discovered AI models are automatically checked against your organization's AI usage policies. If a newly discovered model violates a policy rule:
* A policy violation record is created
* Webhook notifications are dispatched (if configured)
* The violation appears on your Policy Compliance dashboard
## Privacy
Cloud Discovery reads **API call metadata only**:
* Model identifiers
* Caller identities (IAM users, service accounts)
* Timestamps and regions
* API method names
Ayliea **never** accesses:
* Prompt content or model inputs
* Model responses or outputs
* Request/response payloads
* Any other cloud resources (storage, compute, databases)
## Getting Started
Deploy an agent to discover all AI platform usage on your network.
Connect your AWS account to discover Bedrock usage.
Connect your GCP project to discover Vertex AI usage.
# GCP Cloud Logging Integration
Source: https://docs.ayliea.com/integrations/gcp-cloud-logging
Discover AI model usage across your Google Cloud environment by connecting Cloud Logging to Ayliea.
## Overview
The GCP Cloud Logging integration automatically discovers AI model usage in your Google Cloud environment by reading audit logs for Vertex AI API calls. This gives your security team visibility into which AI models are being used, by whom, and how frequently.
## What Data Is Collected
Ayliea reads **API call metadata only** from Cloud Logging:
* **Model names** — which Vertex AI models are invoked (e.g., `gemini-pro`)
* **User identities** — service accounts or users making the calls
* **Timestamps** — when calls occurred
* **Method names** — the API method (e.g., `predict`, `generateContent`)
Ayliea **never** reads prompt content, model responses, or any payload data. Only audit log metadata from Cloud Logging is accessed.
## Prerequisites
* A Google Cloud project with Vertex AI enabled
* Permission to create service accounts and manage IAM roles
* Cloud discovery included in your engagement scope
## Setup Steps
### 1. Create a Service Account
1. Go to the [GCP Console](https://console.cloud.google.com/)
2. Navigate to **IAM & Admin > Service Accounts**
3. Click **Create Service Account**
4. Name it (e.g., `ayliea-logging-reader`)
5. Click **Create and Continue**
### 2. Add the Logging Viewer Role
1. In the **Grant this service account access** step, add the role:
* **Logs Viewer** (`roles/logging.viewer`)
2. Click **Continue**, then **Done**
The Logs Viewer role grants **read-only** access to log entries. It cannot modify any GCP resources.
### 3. Download the JSON Key
1. Open the service account you created
2. Go to the **Keys** tab
3. Click **Add Key > Create new key**
4. Select **JSON** format
5. Download and save the key file
The JSON key file contains credentials. Store it securely and delete it from your local machine after entering it in Ayliea.
### 4. Connect in Ayliea
1. Go to **Organization > Cloud Integrations**
2. Click **Connect** on the GCP Cloud Logging card
3. Paste the contents of your service account JSON key file
4. Click **Test Connection** to validate
5. Click **Connect** to save
## Polling Schedule
Once connected, Ayliea polls Cloud Logging **daily at 7:00 AM UTC**. Each poll covers the time since the last successful poll. The first poll covers the previous 24 hours.
## Troubleshooting
### "Invalid service account JSON format"
* Ensure you pasted the **entire** JSON key file contents, including the opening and closing braces
* Verify the JSON contains `client_email`, `private_key`, and `project_id` fields
### "Invalid credentials or insufficient permissions"
* Confirm the service account has the `roles/logging.viewer` role
* Check that the service account key hasn't been revoked
* Verify Cloud Logging API is enabled in your project
### "Failed to connect to GCP Cloud Logging"
* Verify the project ID in the service account JSON is correct
* Check that the Cloud Logging API is enabled for your project
### No platforms discovered after connecting
* Confirm that Vertex AI is being used in the project
* Audit logs may take a few minutes to appear in Cloud Logging
* The first poll covers only the last 24 hours — wait for the next daily poll
## Security Considerations
* Credentials are encrypted at rest
* The Logs Viewer role grants **read-only** access to log entries only
* Ayliea never accesses Cloud Storage, Compute Engine, or any other GCP services
* You can revoke access at any time by deleting the service account key or disconnecting in Ayliea
# Network Collector
Source: https://docs.ayliea.com/integrations/network-collector
Deploy the Ayliea Collector to discover AI platform usage through DNS and TLS traffic analysis.
## Overview
The Ayliea Collector is a lightweight network agent that passively discovers AI platform usage on your network. It captures DNS queries and TLS Server Name Indication (SNI) handshakes — both plaintext metadata — and forwards matched events to Ayliea for analysis.
**Privacy-first by design:**
* No SSL/TLS decryption
* No packet payload inspection
* No content analysis
* No endpoint agents required
The collector ships with a built-in registry of 57+ AI platforms and 177+ domains, covering LLM providers, code assistants, image generators, and enterprise AI services.
## Deployment Methods
Choose the method that fits your environment:
Recommended for most deployments. Hardened container with minimal privileges.
Self-contained binary for hosts without Docker. Linux x86\_64 and ARM64.
Point your existing firewall syslog at Ayliea. No agent required.
## Prerequisites
Before deploying, generate a monitoring token in Ayliea:
Go to **Discovery** in the left sidebar of your Ayliea dashboard.
In the **Monitoring Tokens** section, click **Generate Token** and enter a descriptive label (e.g., "prod-firewall" or "hq-network").
Copy the token immediately — it is shown only once. You will need it for all deployment methods.
Monitoring tokens are shown once at creation. If you lose a token, revoke it and generate a new one.
## Docker
The recommended deployment method. The container runs as a non-root user with a read-only filesystem and minimal Linux capabilities.
Save the following as `docker-compose.collector.yml`:
```yaml docker-compose.collector.yml theme={null}
services:
ayliea-collector:
image: ghcr.io/ayliea/ayliea-collector:latest
container_name: ayliea-collector
network_mode: host
cap_drop:
- ALL
cap_add:
- NET_RAW
security_opt:
- no-new-privileges:true
read_only: true
tmpfs:
- /tmp:size=64m,noexec,nosuid
- /run:size=16m,noexec,nosuid
deploy:
resources:
limits:
memory: 256m
cpus: "1.0"
environment:
- AYLIEA_TOKEN=YOUR_TOKEN_HERE
- AYLIEA_ENDPOINT=ingest.ayliea.com
- INTERFACE=eth0
- SYSLOG_PORT=6514
- SYSLOG_PROTOCOL=tls
- HTTPS_ENABLED=true
- HEALTH_PORT=9090
- LOG_LEVEL=INFO
restart: unless-stopped
logging:
driver: json-file
options:
max-size: "10m"
max-file: "3"
```
Replace `YOUR_TOKEN_HERE` with your monitoring token and `eth0` with your network interface.
```bash theme={null}
ip link show
```
Common values: `eth0` (Linux), `ens33` (VMware), `eno1` (Dell/HP servers), `bond0` (bonded interfaces).
```bash theme={null}
docker compose -f docker-compose.collector.yml up -d
```
```bash theme={null}
curl -s http://127.0.0.1:9090/health | python3 -m json.tool
```
You should see `"status": "running"` and `events_captured` incrementing as traffic flows.
If port 9090 is already in use on your host, change the `HEALTH_PORT` environment variable to an available port.
### Security hardening
The Docker configuration above includes production-grade security defaults:
| Setting | Purpose |
| - | - |
| `cap_drop: ALL` + `cap_add: NET_RAW` | Only the packet capture capability is granted |
| `read_only: true` | Container filesystem cannot be modified |
| `no-new-privileges` | Prevents privilege escalation inside the container |
| `tmpfs` with `noexec,nosuid` | Temp directories cannot execute binaries |
| `memory: 256m`, `cpus: 1.0` | Limits resource consumption |
| Non-root user | Container process does not run as root |
### Environment variables
| Variable | Default | Description |
| - | - | - |
| `AYLIEA_TOKEN` | *(required)* | Your organization's monitoring token |
| `AYLIEA_ENDPOINT` | `ingest.ayliea.com` | Ingest endpoint hostname |
| `INTERFACE` | `eth0` | Network interface to capture on |
| `SYSLOG_PORT` | `6514` | Syslog forwarding port |
| `SYSLOG_PROTOCOL` | `tls` | Transport protocol: `tls`, `tcp`, or `udp` |
| `HTTPS_ENABLED` | `true` | Enable HTTPS batch forwarding |
| `HEALTH_PORT` | `9090` | Local health endpoint port |
| `LOG_LEVEL` | `INFO` | Log verbosity: `DEBUG`, `INFO`, `WARNING`, `ERROR` |
## Static Binary
A self-contained binary for Linux hosts without Docker. No Python runtime required — only `libpcap`.
Choose your architecture:
| Platform | Binary |
| - | - |
| Linux x86\_64 | `ayliea-collector-linux-amd64` |
| Linux ARM64 | `ayliea-collector-linux-arm64` |
Run `uname -m` to check: `x86_64` = amd64, `aarch64` = arm64.
```bash theme={null}
# Download binary and checksum
curl -LO https://github.com/Ayliea/ayliea/releases/latest/download/ayliea-collector-linux-amd64
curl -LO https://github.com/Ayliea/ayliea/releases/latest/download/ayliea-collector-linux-amd64.sha256
# Verify integrity
sha256sum -c ayliea-collector-linux-amd64.sha256
# Make executable
chmod +x ayliea-collector-linux-amd64
```
```bash theme={null}
# Debian/Ubuntu
apt install libpcap0.8
# RHEL/CentOS/Fedora
yum install libpcap
```
Most server distributions include `libpcap` by default.
```bash theme={null}
sudo ./ayliea-collector-linux-amd64 \
--token YOUR_TOKEN \
--interface eth0
```
The binary requires root or `CAP_NET_RAW` for packet capture.
### CLI options
All options can also be set via environment variables. CLI flags take precedence.
```
ayliea-collector [OPTIONS]
Options:
--token TOKEN Ingest token (env: AYLIEA_TOKEN)
--interface IFACE Network interface (env: INTERFACE, default: eth0)
--endpoint HOST Ingest hostname (env: AYLIEA_ENDPOINT, default: ingest.ayliea.com)
--syslog-port PORT Syslog port (env: SYSLOG_PORT, default: 6514)
--syslog-protocol PROTO tls | tcp | udp (env: SYSLOG_PROTOCOL, default: tls)
--no-https Disable HTTPS batch forwarding (env: HTTPS_ENABLED=false)
--log-level LEVEL DEBUG | INFO | WARNING | ERROR (env: LOG_LEVEL, default: INFO)
```
### Running as a systemd service
For persistent deployments, create a systemd unit:
```ini /etc/systemd/system/ayliea-collector.service theme={null}
[Unit]
Description=Ayliea Collector
After=network-online.target
Wants=network-online.target
[Service]
Type=simple
ExecStart=/usr/local/bin/ayliea-collector --token YOUR_TOKEN --interface eth0
Restart=on-failure
RestartSec=10
AmbientCapabilities=CAP_NET_RAW
NoNewPrivileges=true
ProtectSystem=strict
ProtectHome=true
User=ayliea-collector
[Install]
WantedBy=multi-user.target
```
```bash theme={null}
# Create service user
sudo useradd -r -s /usr/sbin/nologin ayliea-collector
# Copy binary
sudo cp ayliea-collector-linux-amd64 /usr/local/bin/ayliea-collector
# Enable and start
sudo systemctl daemon-reload
sudo systemctl enable --now ayliea-collector
```
## Syslog Forwarding
If your firewall or proxy already exports syslog, you can point it directly at Ayliea without deploying an agent.
### Ingest endpoints
| Protocol | Endpoint | Port |
| - | - | - |
| TLS (recommended) | `ingest.ayliea.com` | 6514 |
| TCP | `ingest.ayliea.com` | 514 |
| UDP | `ingest.ayliea.com` | 514 |
### Authentication
Include your monitoring token in the RFC 5424 structured data element on every syslog message:
```
[ayliea@00000 key="YOUR_TOKEN"]
```
Most enterprise firewalls (Palo Alto, Fortinet, pfSense) support custom structured data in their syslog forwarding configuration.
### Supported log formats
| Vendor | Format | Notes |
| - | - | - |
| Palo Alto Networks | CSV traffic/URL filtering logs | Auto-detected |
| FortiGate | Key=value syslog or CSV export | Auto-detected |
| pfSense / OPNsense | CSV export | Auto-detected |
| Squid Proxy | Native and CLF formats | Auto-detected |
| Windows DNS Server | Debug log format | Auto-detected |
| BIND | Query log format | Auto-detected |
| Zscaler | CSV web log exports | Auto-detected |
| Generic CSV | Any CSV with domain/IP/timestamp columns | Fallback parser |
## Health Endpoint
Both the Docker container and the binary expose a local health endpoint (default: `http://127.0.0.1:9090/health`):
```json theme={null}
{
"status": "running",
"uptime_seconds": 3600,
"interface": "eth0",
"events_captured": 1542,
"events_matched": 87,
"platforms_detected": 12,
"syslog_connected": true,
"https_connected": true,
"last_event_at": "2026-04-10T21:15:48.230224+00:00"
}
```
Use this endpoint for monitoring and integration with your existing health check tooling (Uptime Kuma, Prometheus, etc.).
## Scan Frequency
The collector buffers captured events and processes them into discovery scans on a configurable schedule. You can set the scan frequency in your Ayliea dashboard under **Discovery > Scan Frequency**.
| Frequency | Description |
| - | - |
| **Hourly** | Events processed every hour. Best for high-security environments needing rapid detection. |
| **Daily** (default) | Events processed once per day. Recommended for most organizations. |
| **Weekly** | Events processed once per week. Suitable for low-traffic or cost-sensitive deployments. |
Regardless of the schedule, events are processed immediately if the buffer reaches 5,000 events.
Events are persisted across collector restarts and deployments — no data is lost during planned maintenance.
## Troubleshooting
### No events captured
* Verify the interface name matches your primary network interface (`ip link show`)
* Confirm the container has `NET_RAW` capability: `docker inspect ayliea-collector | grep CapAdd`
* Check that DNS and TLS traffic is actually flowing on the specified interface
* Set `LOG_LEVEL=DEBUG` for verbose capture output
### Syslog forwarding errors
* Confirm `ingest.ayliea.com` resolves from the host: `dig ingest.ayliea.com`
* Check that port 6514 (TLS) or 514 (TCP/UDP) is not blocked by a firewall
* Verify your token is valid and has not been revoked
* Check your firewall's syslog forwarding logs for transmission errors
### Health endpoint unreachable
* The health endpoint binds to `127.0.0.1` — it is only accessible from the host itself
* If another service uses port 9090, set `HEALTH_PORT` to an unused port
### Connection verification timeout
After deploying, use the **Verify** button in the Ayliea dashboard to confirm events are flowing. If verification times out after 5 minutes:
* Generate test traffic by visiting an AI platform from a device on the monitored network (e.g., `curl https://api.openai.com`)
* Check the health endpoint for `events_captured > 0`
* Verify the token matches exactly (no extra spaces or quotes)
# Introduction
Source: https://docs.ayliea.com/introduction
How Ayliea's scoped security assessments, free readiness quiz, and engagement portal fit together.
Ayliea conducts independent, scoped AI security assessments and cybersecurity risk reviews. An engagement starts with the system, decision, and evidence that need review. The written scope defines the procedures and deliverables; findings identify the evidence considered, gaps, limitations, and prioritized actions. See [assessment services](https://ayliea.com/services) or [Oklahoma City cybersecurity consulting](https://ayliea.com/oklahoma-city-cybersecurity-consulting) for the current public offers.
This documentation explains the Ayliea client portal and its assessment workflows. Portal access and available features depend on the agreed engagement. Ayliea provisions an organization and invites its users; there is no public portal sign-up.
The [free AI readiness quiz](https://ayliea.com/assess) is a 14-question, self-reported starting point that runs without an account. It is separate from a scoped, evidence-based assessment and does not produce an independent assurance conclusion.
## Who It's For
Review the agreed controls, evidence gaps, and remediation priorities in the context of an engagement.
Document the in-scope AI systems, data flows, vendor dependencies, and safeguards that matter to the review.
Understand the assessor's findings, limitations, and next actions before making a risk decision.
## In the Client Portal
After Ayliea provisions your organization, the portal can support assessment questions, evidence references, scores, recommendations, reports, and deliverables included in your scope. A self-reported score describes the answers entered into the portal. The assessor's final report states what evidence was reviewed and what could or could not be established.
Relevant findings may be mapped to frameworks such as NIST AI RMF or NIST CSF. Framework mapping is not certification, a legal opinion, or proof that every control was tested. See [Getting Access](/getting-started/creating-account) and the [Quickstart](/getting-started/quickstart) for the portal workflow.
## Pricing and Scope
[Current assessment options and starting prices](https://ayliea.com/pricing) are maintained on the website. The scoping conversation determines the actual systems, evidence, procedures, exclusions, timeline, and final fee. Portal access is governed by the engagement, not a required public software subscription.
## Get Started
Get a self-reported AI readiness snapshot without creating an account.
Discuss the review need and agree on scope before assessment work begins.
Follow the invitation sent after Ayliea provisions your organization.
Learn how the portal organizes its assessment content and references.
# Assessments
Source: https://docs.ayliea.com/product/assessments
How to start, resume, and complete security assessments
An assessment is a structured pass through all categories and questions in a compliance framework. Your answers are saved automatically, and Ayliea calculates a weighted score when you finish every category.
## Starting a New Assessment
Tap the **Assess** icon in the bottom navigation bar. The framework picker loads automatically.
Each framework card shows the framework name, a brief description, category count, and total question count. Tap the card for the framework you want to assess against.
Free assessments are limited to **AI Security** (AISS). All other frameworks are available once your organization has an active engagement. See [Frameworks Overview](/frameworks/overview) for a comparison.
After selecting a framework, the category list loads. Tap any category to start answering its questions. You do not need to complete categories in order.
## Question Categories
Each framework divides its control set into named domains called categories. Categories map to the compliance domain structure defined by the framework — for example, CIS Controls v8.1 groups questions by individual CIS Controls (Control 1: Inventory of Enterprise Assets, Control 2: Inventory and Control of Software Assets, and so on), while NIST CSF 2.0 organizes questions by its six core Functions (Govern, Identify, Protect, Detect, Respond, Recover).
Each category card on the Assess tab shows:
* The category name and icon
* Total number of questions in the category
* A completion badge once all questions are answered
## Answering Questions
Questions are presented one at a time within a category. A progress bar and question navigator at the top of the screen show your position within the category and let you jump to any question directly.
Ayliea uses four question types depending on the control being evaluated. For full detail on how each type works and how points are awarded, see [Your First Assessment](/getting-started/your-first-assessment).
A binary compliance check — full points for Yes, zero for No.
Select the one option that best describes your implementation maturity. Each option carries a different point value.
Select all applicable sub-controls. Partial credit is awarded proportionally based on how many items are checked.
Rate your maturity on a 1–5 scale. Higher values represent more documented, tested, and continuously monitored practices.
Every question has an optional **Notes** field for recording evidence references, ticket numbers, or implementation caveats.
## Progress Tracking
Progress is tracked at two levels:
**Category level** — the category list displays a completion badge on each category you have fully answered. A progress count (e.g., *12 / 18 answered*) appears on in-progress categories.
**Framework level** — the Assess tab header shows the total number of questions answered across all categories in the selected framework.
Both progress indicators update in real time as you answer questions.
## Resuming In-Progress Assessments
Answers save automatically each time you select a response. If you close the app or navigate away mid-category, your progress is preserved. When you return to the Assess tab and select the same framework, the category list reflects your saved answers and you can continue from where you left off.
There is no explicit "save" button — every answer is written to the database as soon as you select it. You will never lose progress from closing the app unexpectedly.
## Multi-Framework Support
You can run independent assessments against different frameworks. Each framework maintains its own assessment record, its own set of answers, and its own score. Switching frameworks in the Assess tab loads that framework's assessment state without affecting others.
This lets you, for example, complete a CIS Controls v8.1 assessment for a baseline posture score and then run a HIPAA assessment to evaluate regulatory readiness — both scores coexist on the Dashboard and in the Results screen.
Starting a new assessment for a framework where you already have a completed assessment will create a fresh record. Your prior completed assessment and its score remain accessible in your assessment history.
# Billing & Invoices
Source: https://docs.ayliea.com/product/billing
How engagement and Monitoring Retainer invoices appear and get paid in your portal
Invoices for your engagement and any Monitoring Retainer are delivered through your portal rather than a separate billing system. The **Invoices** page shows every invoice your assessor has issued to your organization.
## What you'll see
Each invoice shows:
* A short description of what it's for
* **Status** — Open, Paid, Void, Uncollectible, or Past due (shown automatically once an open invoice passes its due date)
* **Amount**, invoice number, and the date it was **issued**
* **Due date** (for open invoices) or **paid date** (for paid invoices)
## Paying an invoice
When an invoice is open, a **Pay** button takes you to a secure, hosted payment page to pay online. Invoices that are void, uncollectible, or already paid don't show a pay link.
Not every organization has invoices to pay through the portal today — some engagements are still invoiced and paid through other arrangements agreed with your assessor. If you're unsure how your engagement is billed, contact your Ayliea engagement lead.
## No invoices yet?
If nothing has been issued to your organization yet, the page shows an empty state: your assessor will send invoices here, and you'll be able to pay online as soon as one is issued.
# Dashboard
Source: https://docs.ayliea.com/product/dashboard
Understanding your security posture dashboard
The Dashboard is the home screen of Ayliea. It gives you an at-a-glance view of your current security posture based on your most recently completed assessment.
## Score Gauge
The score gauge is a circular arc in the center of the dashboard. It displays three pieces of information:
* **Numeric score** — your overall weighted score from 0 to 100
* **Letter grade** — A through F derived from your numeric score
* **Grade label** — a plain-language descriptor (Excellent, Good, Fair, Poor, or Critical)
The arc color changes with your score — green for high scores, red for critical scores — so you can read your posture at a glance without focusing on the number.
The gauge animates from zero each time it loads, counting up to your current score.
The score displayed reflects your **most recently completed** assessment for the active framework. If you have completed assessments across multiple frameworks, use the framework selector in the Assess tab to switch context.
## Category Breakdown
Below the score gauge, the Category Breakdown shows per-category scores as labeled horizontal progress bars. Key behaviors:
* **Weakest-first sort** — categories are ordered by score ascending, so the areas needing the most attention appear at the top
* **Color coding** — each bar uses the same color scale as the gauge (red for critical, green for strong)
* **Collapsed by default** — if the framework has more than 6 categories, only the 6 weakest are shown initially. Tap **Show all** to expand the full list
* **Desktop layout** — on wider screens, the breakdown renders in two columns to reduce vertical scrolling
Each bar shows the category name on the left and the numeric percentage on the right. Tapping a category row navigates to that category's question list so you can review or update your answers.
## Framework Badge
When a completed assessment is shown, a framework badge appears indicating which standard the score belongs to — for example, **CIS Controls v8.1** or **NIST CSF 2.0**. This badge helps you identify which assessment you are viewing if you have completed assessments across multiple frameworks.
## Score Trends
Score history over time (a chart comparing successive assessment runs) is a planned feature and is not yet available. Once you complete multiple assessments for the same framework, trend data will appear here automatically.
## Empty State
If you have not yet completed any assessment, the dashboard shows an empty state with a prompt to start your first assessment. The score gauge and category breakdown are not shown until at least one assessment is complete.
Complete the [AI Security](/frameworks/ai-security) framework first — it's free, no engagement required, and takes roughly 20–30 minutes. Your dashboard will populate as soon as you finish all categories.
# Deliverables
Source: https://docs.ayliea.com/product/deliverables
Find the files and reports your assessor publishes to your portal as your engagement progresses
As your engagement progresses, your assessor publishes files to your portal — consulting artifacts and report snapshots that make up your signed deliverable package. The **Deliverables** page in your portal is where you find and download them.
## What you'll see
Deliverables are grouped by engagement. Each engagement group shows:
* **Package** — which engagement (Focused, Comprehensive, or Enterprise) the deliverables belong to
* **Delivered badge** — appears once your assessor has marked the engagement complete
* A table listing each file's **title**, **type** (Consulting Artifact or Report Snapshot), the date it was **published**, its **size**, and a **Download** link
If you're in the middle of an active engagement, you may not see anything here yet — your assessor publishes files as work is completed and reviewed, not continuously.
## Live results vs. deliverables
Deliverables are the signed, point-in-time package your assessor publishes. Your live, continuously-updated scores, assessment answers, and recommendations are always available from your [Dashboard](/product/dashboard) and assessment results — the Deliverables page links directly to both so you can move between the two.
## No deliverables yet?
If your engagement hasn't reached a delivery milestone, the page shows an empty state: your assessor will publish files and reports here once they're ready. If you believe something is missing, contact your Ayliea engagement lead.
# Organizations
Source: https://docs.ayliea.com/product/organizations
Manage multi-seat access, invite team members, assign roles, and collaborate on assessments and remediation within your Ayliea organization.
An Ayliea organization is a shared workspace that lets multiple team members collaborate on assessments, share results, divide remediation work, and generate reports together. Every engagement client works within an organization.
## Creating an Organization
Your organization is created as part of getting access to Ayliea — see [Getting Access](/getting-started/creating-account) for how account and organization provisioning works. The person your team designates during scoping becomes the Owner.
If your organization already exists and you have been invited to join, you do not need to create one — accept the invitation link from your email to join automatically after signing in.
## Roles
Organization members have one of three roles that determine what they can do within the workspace.
| Role | Permissions |
| - | - |
| **Owner** | Full access to all features, billing, organization settings, and member management. There is always exactly one Owner per organization. |
| **Admin** | Can manage members, assign roles, run assessments, view all results, and generate reports. Cannot change billing. |
| **Member** | Can run assessments, view shared results, and update recommendation status. Cannot manage members or billing. |
Roles are assigned when sending an invitation and can be changed at any time by an Owner or Admin.
## Inviting Team Members
To invite someone to your organization:
1. Navigate to **Organization** from the main navigation
2. Open the **Members** section and select **Invite Member**
3. Enter the invitee's email address and select their role
4. Click **Send Invitation**
The invitee receives an email with a secure link. After clicking the link and completing sign-up (or signing in if they already have an account), they are automatically added to your organization with the assigned role.
Pending invitations are visible in the Members section and can be cancelled before they are accepted.
## Managing Members
Owners and Admins can view all organization members, change member roles, and remove members from the organization. Member changes take effect immediately.
When a member is removed, their assessment contributions and assigned recommendations remain visible to the organization. Only their active access is revoked.
## Shared Assessments
Within an organization, assessments and their results are visible to all members. This means:
* Any member can see assessment scores, category breakdowns, and recommendation lists
* Recommendations can be assigned to specific members for ownership
* Report generation is available to all members
There is no concept of a private assessment within an organization workspace — all completed assessments are shared by default.
## External Auditor Access
To share results with an external auditor, use [Report Sharing](/product/reports#sharing-with-auditors) rather than adding the auditor as an organization member. Auditor share links provide scoped, time-limited access to specific reports without granting any access to your live workspace or assessment data.
## Organization Settings
Owners can configure the following from Organization Settings:
* **Organization name** — displayed on reports and in the workspace
* **Custom branding** — upload a logo for use on generated PDF reports
* **Member management** — invite, remove, and change roles
Billing is handled by the Owner account — see [Billing & Invoices](/product/billing). Admins and Members cannot view or change billing details.
# Accessing Your Portal
Source: https://docs.ayliea.com/product/portal
Sign in to your Ayliea portal and find your way around your workspace
Your Ayliea portal is your organization's secure workspace — where you run assessments, review scores and recommendations, and (once your engagement reaches those stages) find your deliverables and invoices. It's part of your workspace once your account is activated; see [Getting Access](/getting-started/creating-account) if you haven't activated your account yet.
## Signing in
Go to your Ayliea portal URL and sign in with your email, password, and a code from your authenticator app (multi-factor authentication is part of every account). If you're signed out, you're redirected back to sign-in and returned to the page you were on afterward.
## What's in your portal
Your overall posture score, category breakdown, and a snapshot of where to focus next.
Run and resume framework assessments, review category scores, and drill into individual questions.
AI-personalized remediation guidance tied to your weakest controls, with playbooks and evidence tracking.
Generate and share branded PDF reports summarizing your posture and progress.
Files and reports your assessor publishes as your engagement progresses.
Engagement and Monitoring Retainer invoices, with online payment when one is issued.
If your organization has an AI Governance engagement, a **Governance** section also appears in the main navigation — see [AI Governance](/governance/overview).
## Organization workspace
If you're part of a multi-seat organization, your portal is shared with your teammates — assessments, scores, and recommendations are visible to everyone in your organization by default. See [Organizations](/product/organizations) for roles and how team access works.
# Recommendations
Source: https://docs.ayliea.com/product/recommendations
How Ayliea generates, prioritizes, and tracks AI-powered remediation recommendations from your assessment results.
Recommendations are AI-generated remediation actions produced after you complete an assessment. Each recommendation targets a specific control gap identified by your answers, and the full list is ranked so that the highest-impact items appear first.
## How Recommendations Are Generated
When you complete all categories in a framework, Ayliea analyzes your answers and generates a prioritized list of recommendations. Each recommendation addresses a specific question where your response indicated a gap — a missing control, an incomplete implementation, or a practice that is not yet documented or tested.
Recommendations are ranked by impact, which takes into account:
* **Category weight** — gaps in higher-weight categories produce recommendations with greater impact on your overall score
* **Point gap** — the difference between what you earned and the maximum possible for that question
* **Control criticality** — certain controls are flagged as high-priority by the framework publisher
The result is a list where fixing the top item moves your score more than fixing any other single item.
## Recommendation Detail
Each recommendation includes:
* **Title** — the specific control or practice being addressed
* **Framework mapping** — which framework category and control the recommendation belongs to
* **Priority level** — Critical, High, Medium, or Low, based on impact rank
* **Implementation guidance** — step-by-step remediation playbook describing how to implement the control
* **Effort estimate** — a rough estimate of the implementation complexity
## Filtering and Navigation
The Recommendations screen supports filtering by:
* **Priority** — view Critical or High items only to focus your immediate effort
* **Category** — filter to recommendations within a specific control domain
* **Status** — filter to open, in progress, or completed items
Use the search bar to find recommendations by keyword if you are looking for a specific control.
## Tracking Progress
Each recommendation has a status that you update as you work through your remediation backlog:
| Status | Meaning |
| - | - |
| Open | Not yet started |
| In Progress | Work has begun |
| Completed | Control implemented |
| Accepted Risk | Control acknowledged but not implemented; risk formally accepted |
Updating a recommendation's status helps your team track what has been done and what remains. Completed and accepted recommendations are excluded from the prioritized list by default.
## Assigning Recommendations
On Organization accounts, recommendations can be assigned to specific team members. Assignments are visible to all members of the organization, making it easy to divide remediation work across your team without losing track of ownership.
## Notes and Evidence
Each recommendation has a notes field where you can record implementation decisions, link to tickets or change requests, or note caveats about your implementation approach. Notes are visible to all organization members with access to the assessment.
After implementing a recommendation, re-answer the associated question in the assessment to see your score update in real time. Recommendations marked Completed that still have the original low-scoring answer will not improve your score until the assessment answer is updated.
## Cross-Framework Recommendations
If you have completed multiple framework assessments, Ayliea surfaces recommendations that address gaps appearing across more than one framework. Fixing a single control gap that appears in both your NIST CSF and ISO 27001 assessments improves both scores simultaneously.
# Reports
Source: https://docs.ayliea.com/product/reports
Generate PDF posture reports combining your assessment results and network discovery data for auditors, executives, and internal reviews.
Ayliea generates PDF posture reports that summarize your security posture in a format designed for sharing outside your team — with auditors, executives, board members, or enterprise customers evaluating your security program.
## What's in a Report
A posture report is a 7-page branded PDF that combines your assessment results with any network discovery data from your account. The report is structured so that a non-technical reader can understand your security posture without needing access to the platform.
| Section | Contents |
| - | - |
| Executive Summary | Overall score, letter grade, trend direction, and a plain-language summary of your strongest and weakest areas |
| Assessment Results | Category-by-category score breakdown with color-coded status indicators |
| Trust Gap | Side-by-side comparison of your self-reported score and your network-observed posture (if discovery data is available) |
| AI Tool Inventory | List of discovered AI tools with risk classification and approval status |
| Remediation Roadmap | Top prioritized recommendations with effort estimates and control references |
| Framework Reference | The compliance framework assessed, question count, and assessment date |
| Methodology | Plain-language explanation of how scores are calculated and what the grade scale means |
## Generating a Report
To generate a report:
1. Navigate to the **Reports** tab from the main navigation
2. Select the assessment you want to report on
3. Choose the report scope — full report or sections only
4. Click **Generate Report**
Report generation takes a few seconds. Once complete, the report is available to download as a PDF or share via a secure link.
## Sharing with Auditors
You can share a report with an external auditor without giving them access to your Ayliea account. From the Reports tab, select a generated report and choose **Share with Auditor**.
Auditor share links:
* Require email verification before the report is accessible
* Expire after a configurable time window (default: 30 days)
* Can be revoked at any time from the Reports tab
* Show only the sections you explicitly include in the share
## Report Scope Controls
Before generating or sharing a report, you can configure which sections to include. This is useful when sharing with different audiences — an executive summary for the board, a full technical report for your auditor, or just the remediation roadmap for your engineering team.
## Branding
Reports display your organization name and the Ayliea logo. Custom branding (your own logo and color scheme) is available on Organization accounts — configure it from your organization settings before generating the report.
## Report History
All generated reports are saved and accessible from the Reports tab. Previous reports are useful for demonstrating posture improvement over time — you can share a report from six months ago alongside a current report to show the progress your team has made.
Reports reflect your posture at the time of generation. If you update assessment answers or receive new discovery data after generating a report, generate a new report to reflect the changes. Previously generated reports are not updated retroactively.
# Scoring
Source: https://docs.ayliea.com/product/scoring
How Ayliea calculates your security posture score
Ayliea produces a numeric score from 0 to 100 for each completed assessment. This score reflects how completely your organization has implemented the controls defined by the selected compliance framework.
## Scoring Methodology
Scores are calculated in two stages:
**Stage 1 — Category scores.** For each category, Ayliea sums the points you earned across all questions and divides by the maximum possible points for that category:
```
category_score = (points_earned / max_points) × 100
```
Each question defines its own maximum point value, and partial credit is available for checklist and scale question types. Points earned are capped at each question's maximum regardless of the response selected.
**Stage 2 — Weighted overall score.** The category scores are combined into a single overall score using a weighted average. Each category carries a weight defined by the framework:
```
overall_score = Σ(category_score × category_weight) / Σ(category_weight)
```
This means categories that cover more critical or broadly applicable controls contribute more to the final score than narrower or lower-priority categories. The exact weights are defined per framework by the respective standard's guidance on control priority.
## Grade Scale
Your numeric score maps to a letter grade as follows:
| Grade | Score Range | Label |
| - | - | - |
| A | 90 – 100 | Excellent |
| B | 75 – 89 | Good |
| C | 60 – 74 | Fair |
| D | 40 – 59 | Poor |
| F | 0 – 39 | Critical |
The grade boundaries are intentionally strict. A **B** (Good) requires 75 or above, reflecting that security posture below 75% leaves meaningful gaps in control coverage. An **A** (Excellent) requires 90 or above and represents a mature, broadly implemented control set.
Both the numeric score and letter grade are displayed on the score gauge on your Dashboard and Results screen.
## Category Weighting
Category weights are defined per framework and are set to reflect the relative importance of each control domain within the standard:
* A category with weight **2** contributes twice as much to the overall score as a category with weight **1**
* Categories with a weight of **0** are informational only and do not affect the overall score
* If all categories carry equal weight, the overall score is a simple average of the category scores
You can see the relative contribution of each category in the [Category Breakdown](/product/dashboard#category-breakdown) on your dashboard. Categories with higher weights that score poorly have a disproportionate impact on your overall score — improving them moves the needle the most.
## Score Range
Scores are expressed as a percentage from **0** (no controls implemented) to **100** (all controls fully implemented). A score of 100 does not indicate zero risk — it reflects full coverage of the assessed framework's control set as evaluated through self-reported answers.
Assessment scores are based on self-reported answers. Ayliea scores your posture against what you report — results are only as accurate as the answers you provide. For controls you are uncertain about, use the **Notes** field to flag items for follow-up verification.
## How to Improve Your Score
Your score improves by implementing controls in your weakest categories. Ayliea generates prioritized AI recommendations after each completed assessment to guide your remediation efforts.
Recommendations are ranked by impact — controls in higher-weight categories with the largest point gaps surface first. Many recommendations include step-by-step implementation playbooks.
See [Recommendations](/product/recommendations) for detail on how to work through your remediation backlog.